The Scientific Study of

Human Nature:
Tribute to Hans J. Eysenck
at Eighty
Titles of related interest

Personality and Individual Differences

The Official Journal of the International Society for the Study of Individual Differences
(ISSID)

Editors-in-Chief:
H. J. Eysenck and S. B. G. Eysenck
Department of Psychology
Institute of Psychiatry
De Crespigny Park
Denmark Hill
London SE5 8AF
UK

Aims and scope

Personality and Individual Differences is devoted to the publication of articles (experi-
mental, theoretical, review) which aim to integrate as far as possible the major factors of
personality with empirical paradigms from experimental, physiological, animal, clinical,
educational, criminological or industrial psychology or to seek an explanation for the
causes and major determinants of individual differences in concepts derived from these
disciplines. The editors are concerned with both genetic and environmental causes, and
they are particularly interested in possible interaction effects. The traditional type of
work on traits, abilities, attitudes, types and other latent structures underlying con-
sistencies in behavior has in recent years been receiving rather short shrift in traditional
journals of personality; Personality and Individual Differences aims to reinstate it to its
proper place in psychology, equal in importance with general experimental work, and
interacting with it to make up a unitary science of psychology.

For further information please contact:

Elsevier Science Elsevier Science
Regional Sales Office Regional Sales Office
Customer Support Department Customer Support Department
P.O. Box 211 655 Ave of the Americas
1000 AE Amsterdam New York, NY 10010
The Netherlands USA
Tel: (+31) 20-485-3757 Telephone: +1-212-633-3730
Fax: (+31) 20-485-3432 Fax: +1-212-633-3680
E-mail: [email protected] E-mail: [email protected]
The Scientific Study of
Human Nature:
Tribute to Hans J. Eysenck
at Eighty

Edited by Helmuth Nyborg

University of Aarhus, Risskov, Denmark

Pergamon
U.K. Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington,
Oxford OX5 1GB, U.K.
U.S.A. Elsevier Science Inc., 655 Avenue of the Americas, New York,
NY 10010, U.S.A.
JAPAN Elsevier Science Japan, Higashi Azabu 1-chome Building 4F, 1-9-15,
Higashi Azabu, Minato-ku, Tokyo 106, Japan

Copyright © 1997 Elsevier Science Ltd

All Rights Reserved. No part of this publication may be reproduced,
stored in a retrieval system or transmitted in any form or by any
means: electronic, electrostatic, magnetic tape, mechanical, photo-
copying, recording or otherwise, without permission in writing from
the publishers.
First edition 1997
Library of Congress Cataloging in Publication Data
The scientific study of human nature: tribute to Hans J. Eysenck at
eighty/edited by Helmuth Nyborg - 1st ed.
p. cm.
Includes indexes.
ISBN 0-08-042787-1 (hardcover)
1. Psychology. 2. Personality. 3. Intellect. 4. Eysenck, H. J. (Hans
Jurgen), 1916- . I. Nyborg, Helmuth.
BF121.S383 1997
150-dc21 97-15771
CIP

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library.
ISBN 0 08 042787 1

Printed and bound in Great Britain by Biddies Ltd, Guildford and

King's Lynn
Contents

Foreword
/. Gray ix
General introduction: Hans Eysenck—the man, his friends
and this book
H. Nyborg xiv

PART I. PERSONALITY 1
1. The psychobiological basis of personality
M. Zuckerman 3
2. Hans Eysenck's personality dimensions: Their number and nature
C. R. Brand 17
3. Individual differences in reactions to reinforcing stimuli are neither
black nor white: To what extent are they Gray?
A. D. Pickering, P. J. Con, J. H. Powell, V. Kumari, J. C. Thornton,
and J. A. Gray 36
4. Temperament and personality: Eysenck's three superfactors
as related to temperamental dimensions
/. Strelau and B. Zawadzki 68
5. Psychology and medicine
D. K. B. Nias 92
6. Psychoticism as a dimension of personality
S. Eysenck 109
7. Classical conditioning, arousal, and crime: A biosocial perspective
A. Raine 122
8. Crime and personality
G. H. Gudjonsson 142
9. Sex and personality
G. D. Wilson 165
10. Extraversion and impulsivity: The lost dimension?
W. Revelle 189
vi The Scientific Study of Human Nature

PART II. INTELLIGENCE 213

Introduction: Hans Eysenck and the study of intelligence
A. R. Jensen 214
11. The psychometrics of intelligence
A. R. Jensen 221
12. Behavioral genetic and biological approaches to intelligence
P. A. Vernon 240
13. Geographical variation in intelligence
R Lynn 259
14. Intelligence and information processing
/. /. Deary 282
15. Malleability and change in intelligence
N. Brody 311

PART III. FURTHER EYSENCKIAN INTERESTS 331

Introduction: Hans Eysenck—A man of many talents
H. B. Gibson 332
16. Classical conditioning and the role of personality
/. Martin 339
17. Eysenck's contribution to understanding psychopathology
G. Claridge 364
18. The psychophysics and psychophysiology of extraversion
and arousal
R. M. Stelmack 388
19. (Im)pure genius—psychoticism, intelligence, and creativity
/. P. Rushton 404
20. Molecular creativity, genius and madness
H. Nyborg 422
21. Eysenck's personality theory and organizational psychology
A. Fumham 462
22. Bursts of creativity and aberrant sunspot cycles: hypothetical
covariations
5. Ertel 491
23. Graphology, astrology, and parapsychology
G. A. Dean, D. K B. Nias, and C. C. French 511
 Contents vii

24. Eysenck as teacher and mentor

A. R. Jensen 543

PART IV. EPILOG 561

25. Psychology as science
H. Nyborg 563
Author index 591
Subject index 609
This page intentionally left blank
Contributors

Chris R. Brand Christopher C. French

Department of Psychology Department of Psychology
University of Edinburgh Goldsmiths College,
Edinburgh, U.K. University of London, U.K.

Nathan Brody Adrian Furnham

Department of Psychology Department of Psychology
Wesleyan University University College London
Middletown, U.S.A. London, U.K.

Gordon Claridge H. B. Gibson

Department of Experimental Psychology 10 Manhattan Drive,
Oxford University Cambridge, U.K.
Oxford, U.K.
Jeffrey A. Gray
Philip J. Corr Department of Psychology
Department of Psychology Institute of Psychiatry
Goldsmiths College University of London
University of London London, U.K.
London, U.K.
Gisli H. Gudjonsson
Geoffrey A. Dean Institute of Psychiatry
Analogic, Subiaco 6008 University of London
Western Australia London, U.K.

Ian J. Deary Arthur R. Jensen

Department of Psychology Department of Education
Edinburgh University University of California at Berkeley
Edinburgh, U.K. California, U.S.A.

Suitbert Ertel Veena Kumari

Psychologisches Institut Department of Psychology
Universitat Gottingen Institute of Psychiatry
Gottingen, BRD University of London
London, U.K.
Sybil Eysenck
Department of Psychology Richard Lynn
Institute of Psychiatry Department of Psychology
University of London University of Ulster
London, U.K. Londonderry, U.K.
 The Scientific Study of Human Nature

Irene Martin J. Phillipe Rushton

Department of Psychology Department of Psychology
Institute of Psychiatry University of Western Ontario
University of London London, Ontario, Canada
London, U.K.
Robert M. Stelmack
David K. B. Nias Department of Psychology
Department of Psychological Medicine. University of Ottawa
St. Bartholomew's Hospital Ontario, Canada
University of London Jan Strelau
London, U.K. Department of Individual Differences
University of Warsaw
Helmuth Nyborg Warsaw, Poland
Department of Psychology
University of Aarhus Jasper C. Thornton
Aarhus, Denmark Department of Psychology
Institute of Psychiatry
Alan D. Pickering University of London
St. George's Hospital Medical School London, U.K.
University of London Philip A. Vernon
London, U.K. Department of Psychology
University of Western Ontario
Jane H. Powell London, Ontario, Canada
Department of Psychology
Goldsmiths College Glenn D. Wilson
University of London Department of Psychology
London, U.K. Institute of Psychiatry
University of London
Adrian Raine London, U.K.
Department of Psychology
University of Southern California Bogdan Zawadzki
Los Angeles, U.S.A. Department of Individual Differences
University of Warsaw
William Revelle Warsaw, Poland
Department of Psychology Marvin Zuckerman
Northwestern University Department of Psychology
Illinois, U.S.A. University of Delaware
Newark, U.S.A.
Foreword

THE MAN AND HIS WORK: PARADOX AND CONTROVERSY

It is a truism to say of almost anyone that he is a person full of paradoxes, but

one that applies to Hans Eysenck in spades.
When I was a graduate student in Hans' Department (in the early 1960s), I
arranged my first ever tete-a-tete with the great man to discuss some research I
was thinking of doing. My business was over in a quarter of an hour and I
waited to be dismissed. There was a silence; embarrassed, I started talking
again, and he politely listened. This happened over and over again until finally,
two hours after the meeting began, it dawned upon me that it was up to me to
bring it to an end. Only later did I discover that he wasn't simply enjoying my
idle conversation, he just doesn't like telling people to leave: the well-bred
introvert to perfection! (I also discovered that he had written me off as a
typically verbose product of Oxbridge; but I believe he has formed a more
favorable impression since!) Yet this introvert is an inveterate risk-taker—an
extravert trait if there ever was one. Indeed, he simply loves a fight. I had
started studying the biological bases of sex differences in the rat's emotional
behavior for my Ph.D. When I told him about this work, he could not
understand why I found the topic of any interest. A few years later, however, he
was deep into it himself. No, this was not due to my inspiring work. What had
happened in between was the rise of the new wave of feminism: suddenly, every
book and pamphlet was proclaiming, not just the equality, but the identity of
what were still called, thankfully, the two sexes. Now the biological basis of sex
differences could contribute to a major controversy, one moreover in which
Hans could tilt at an emerging orthodoxy, just as he had already tilted at the
existing orthodoxies of psychoanalysis and the authoritarian Left, and would go
on to tilt at the antismoking lobby and political correctness in all its forms. So,
in private, a calm introvert unable to show a logorrhoeic student to the door; in
public, a fighter keen to get into any scrape, the riskier the better.
So how does the balance sheet look, after a life-time of scientific
controversy? There is no doubt he won the battle over psychoanalysis. Even
in the United States, where this pseudoscience once dominated every
department of psychiatry and most departments of psychology, it is now
dead or dying; and Hans, once hated by the American psychiatric
xii The Scientific Study of Human Nature

establishment, is now invited as an honored guest speaker to their conferences.

The model of clinical treatment that he created to replace psychoanalysis,
behavior therapy (now in its newer, cognitive-behavioral guise), has been
exported to the whole world from its cradle in South London. The similarities
in personality that unite Fascists and Communists, once the cause of serious
scandal to the Liberal Left, are commonplace now that Uncle Joe Stalin's
smiling face has been buried under the realities of the Soviet state. So Hans
won that battle too. But that was not the end of his offenses against the
political Left. When Jensen raised the issue of racial differences in IQ, Hans
again could not resist stepping into the fray. If you look carefully at what he
wrote on that issue, all the arguments are as calm and rational as is private
conversation with this introvert—all the right caveats are showcased, the "not
proven" verdicts recorded. But it was enough that he considered a genetic basis
to race differences in IQ as an open issue to make him, once more and even
more, a political pariah (this was the period when he was physically attacked at
that bastion of civilized learning, the London School of Economics). The issue
is still open: racial differences in IQ persist, but their origin is still unexplained.
Soon the genes that determine individual variation in IQ (as in so many other
psychological traits) will be in the test tube and the issue will be resolved, one
way or the other. On another front, however, Hans probably went on for too
long saying an issue is still open: he was surely right to point out, initially, that
correlations between smoking and a variety of diseases do not by themselves
indicate causation; but since then the case against smoking has accumulated
until it is overwhelming, with recent experimental evidence providing at least
one biochemical causal link (hitherto a serious gap) in the chain.
These have been important and tune-consuming issues in Hans' hectic
scientific life; even, one might say, distractions. For there is no doubt where his
major achievement lies: in the theory and evidence that he has put together,
breathtaking in their scope, which make up nine-tenths of what anyone might
mean, in 1997 as at any time since the early 1950s, by the phrase (the title of
one of his most important books), the biological basis of personality. The
theoretical edifice that he has constructed in this area of research, central to so
much of experimental and clinical psychology, not only defines the field today,
it sets both the standard and the very mold of what such theories will look like
for many decades to come. For this work, even if there were no other in his
extraordinary compass, his place in the history of psychology is assured. He
should also have been assured of honors in his lifetime (he is, after all, the most
widely cited living psychologist); but these have been a long time coming. That
is surely due to all the scrapes he has got into! There are simply too many
enemies around, readier to point to the flaws than to the truly solid
achievement, as I have sadly found on several occasions when I have tried to
remedy his deplorable lack of recognition in his own country. But at least the
last decade has seen a dramatic change in the picture in the United States,
 Foreword xiii

where the American Psychological Association has successively honored Hans

with its Distinguished Scientist Award (1988), the Presidential Citation for
Scientific Contribution (1993), the William James Fellow Award (1994), and
the Centennial Award for Distinguished Contributions to Clinical Psychology
(1996).
To these marks of recognition, and to the 1991 Distinguished Contributions
Award of the International Society for the Study of Individual Differences
(which hosted the publication of this volume), we add our Festschrift, counting
ourselves privileged to share in the celebration of Hans' eightieth birthday.

JEFFREY A. GRAY
General introduction: Hans Eysenck—the man,
his friends and this book

H. Nyborg

THE MAN

Hans Eysenck already has a Festschrift: Dimensions of Personality: Papers in

Honour of H.J. Eysenck (Lynn, 1981). Now he gets another one. That is fairly
unusual. But then we are also talking about an unusual man. When I first met
Eysenck many years ago at a conference in Germany, he was already a legend.
Not only had he written more extensively and authoritatively on more different
topics in psychology than anybody else, but he was also generally known as an
overly polemic, razor-sharp, arrogant, and towering figure. By incidence I
came to sit next to the great man at a lecture. We both barely survived a
grandiose and megalomane presentation by one of the then best-known Soviet
proponents of pure and clean social-reductionism—so popular at the time. For
a long time Eysenck posed as a sphinx, as is his habit when attending lectures.
Near the end it all became too much for him. He first leaned towards me and
said with his usual imperturbable voice, "This man is not saying one word." As
the lecture at long last came to an end, he then rose to his feet and launched a
devastating attack, the force and sharpness of which correlated highly
negatively with his softly smiling face, controlled intonation, and reserved
gestures. This is part of the paradox Jeffrey Gray refers to in his foreword:
Eysenck exhibits at one and the same time the prime characteristics of an
introvert and an extravert. Let me add, that his social desirability score cannot
amount to much.
Many years later, Eysenck gave a public lecture while I was working in
Jeffrey's laboratory at Oxford University. Barely had he entered the scene to
deliver his speech before the fire alarm went off and sprayed an infernal noise
over the full auditorium. As connoisseurs would expect, Eysenck just smiled
and waited patiently for a very long time until somebody eventually got to the
system and turned it off. Eysenck went back to the platform. Precisely at that
time, the bell started all over again on its devastating mission. What reaction
would you expect of a low E low N person? Eysenck took retreat to the single
chair left behind on the scene. There he sat down calmly, took out with overly
 General introduction xv

controlled movements a newspaper from his document case, leaned back

comfortably and began to read, with a faint smile on his face, in front of the
large audience. But when the bell was finally put to rest, he delivered a sharp
and unforgiving lecture right to the point, and with no comments whatsoever
about the obvious attack on his academic freedom to speak. In an illustrative
book, Race Differences and Bias in Academe Pearson (1991) reports on several
such incidences, all testifying to the fact that we are talking about an unusual
man with a complex combination of traits and talents, controversial ideas and a
high "emotional stimulus value" to many.
Most of us ordinary people concentrate of sheer necessity on a few fairly
well-defined areas making up our field of specialization. Not so for Hans. His
works through more than half a century reflect a breathtaking range of
interests, spanning over areas each of which have occupied an army of busy
researchers for ages. Moreover, he is never afraid of leaving the safe zone of
well-trodden orthodox avenues. In the style of a full-blown sensation seeker he
never covers an area dispassionately. Quite to the contrary, Eysenck often
engages with a zest that to some has become synonymous with a scrupulous
crusade for a particular view. The only license for indulging in such a luxury is,
of course, to keep both feet firmly on scientific ground. I believe Eysenck
deserves the right to hold such a license. His unfailing insistence on facts at the
behalf of fiction, his uncompromising habit of telling the emperor that he
needs better offers from more trustworthy dressers in the future, and his
repeated call for substituting loose psychology and clever word games with
solid experiments and natural science marks most of his moves. Perhaps for
exactly these reasons it has become almost a mystery to many how easily
Eysenck drifts towards the wilder shores of fringe science, thereby
embarrassing many of his supporters and bringing them to the brink of
disbelief. He has indulged—and still does—in graphology, parapsychology,
astrology, and possible effects of solar activity on human creativity. He
certainly also shocked some traditional supporters when he stated, almost 20
years ago, that : "Were I starting out in psychology again, and I had a free
hand, I think I would probably choose hypnosis as the primary objective of
study—unfortunately we never have a second chance like that!" (Eysenck,
1978). He, nevertheless, has constantly taken third chances by moving
shockingly fast from one topic to an entirely new one, taking new chances,
allowing his inner "Catherine wheel" to spin so fast and sparkle in unexpected
directions, selecting a primary objective for only a while, leaving it, then being
on the road again moving towards new height; or, perhaps, taking it up again
later, expanding, explaining, or denoucing it.
xvi The Scientific Study of Human Nature

HIS FRIENDS

When Sybil and Hans Eysenck and I dined in his favorite restaurant in London
many years ago, Hans leaned over the huge menu card and mumbled
something. "What was that?" I asked. He turned slowly towards me, smiled
and said, "Wer hat wahl, hat quail!" These words of Goethe in German, his
first language, came easily to mind when I faced the task of inviting former
students and friends to contribute to this Festschrift. Hans' active career spans
more than half a century, and some of the authors in Richard Lynn's
Festschrift had retired. One wrote back to me, "Not having the endurance of
Hans, I have retired 100%, so...". So had others of his former students. Luckily
there still were scores of eminent researchers and clinicians to choose from.
Let me here express my gratitude to those who so enthusiastically endorsed the
plan for the Festschrift, accepted to write a chapter with offendingly brief
notice, often having to set aside urgent matters in order to meet the not too
distant deadline.
Their task was defined in accordance with a suggestion by Irene Martin
(thanks Irene!): first, describe Hans Eysenck's contribution to the field
(including theory where relevant); then what research has developed from it,
and what kinds of amendments/modifications/additions to his work (theory)
are appropriate; and, finally, describe your thoughts about the future of the
field.
I further asked Jeffrey Gray, Arthur Jensen, and Tony Gibson to write the
foreword and introductions to part II and III, respectively. In addition to
asking Art Jensen this and to write a chapter on the psychometrics of
intelligence, I further pressed him to collect, in a separate chapter, the vivid
memories he has from the experience with Eysenck as teacher and mentor.
From pleasant interaction with Art I knew that, in addition to mastering the
field of intelligence, he is also an astute observer of people and an excellent
reporter, despite his own admission that he is not the greatest fraternizer of all.
As a postdoc in the department he watched Hans at close quarters for three
years. Chapter 24 witnesses his impressions.
Allow me to sidetrack for a moment. I wish Art would write soon about his
experiences with another giant in science, the Nobel prize winner William
Shockley. Like Eysenck, Shockley was a controversial person, but with a P
score probably somewhat higher than Eysenck's'. When I first met Shockley he
stuck his heavy glasses distressingly close to mine, and wheezed aggressively,
"So, you are one of the idiots believing that socialization explains it all." I was
only too happy to deny this, but he continued, "What do you believe in, then?"
Before I had had a chance to answer, he said, "Well, I have a prostate problem,
so why don't you tell me what you think when you accompany me to the men's
room?" I will spare the reader the details of the ridiculous situation in the
men's room where we spent quite a while, trying to combine the long-term
 General introduction xvii

business of dehydration with an agitated discussion of nature-nurture

problems, to the obvious amusement of other visitors. The only reason for
mentioning this event is that having received the Nobel prize for inventing the
transistor (with two colleagues), Shockley turned his genius full force towards
eugenic matters, behavior genetics and brain science, and sided with Eysenck
and Art Jensen on several controversial nature-nurture matters. However, he
did so in ways that made him look an enfant terrible 10 times Hans and Art in
the eyes of many (Pearson, 1992). All three approach the marks of genius, but
in different ways (for discussions of genius, see chapters 6, 19, and 20), though
Shockley with uncharacteristic modesty once admitted to Art that he only
ranked as a minor genius, if at all. Once, when Hans visited my home in
Denmark, I asked him over breakfast whether he was a genius. Without
hesitation he answered, "No, my P score is too low." After a pause he added,
"My N score is zero!" This is the answer of a scientist. No nonsense.

THIS BOOK

The book comes in four parts: part I concentrates on Eysenck's influence on

personality research, part II on intelligence, part III on further Eysenckian
interests, and the epilog focuses on psychology as a science. The following
introduces the reader to selected details in parts I and II, and very briefly to
parts III and IV.

Part I
Chapter 1 by Marvin Zuckerman deals with personality from a psychobiolo-
gical point of view. After a historical introduction to the evolution of Eysenck's
personality theory, Zuckerman demonstrates its debt to Pavlovian thinking and
illustrates the unfolding of the P, E, and N dimensions. Zuckerman then
focuses on his own related research on sensation seeking, the refinement of
SSS scales, and on the development of the "Alternative-5" model. Zuckerman
then inquires into the genetics and psychopharmacology of the model and asks
pertinent questions about the role of arousal. The chapter concludes with
considerations about future steps in the area, which includes refinement of
genotype-environment correlation analyses, studies on the receptor side of
neurochemical systems and the new prospects in neuroimaging. Zuckerman
assumes that Eysenck's personality theory will provide the tools needed for
linking the genotype to behavior and the environment via physiological,
hormonal, and neurological sets of intermediaries.
Chapter 2 by Chris Brand deals with the still wide-open question about how
many personality dimensions are really needed. In his historical introduction,
Brand inquires into Eysenck's project of going beyond psychoanalytic, social
xviii The Scientific Study of Human Nature

psychological, and cognitive theory, and of turning the study of personality into
a science, envisaging surface differences to deeper, underlying dimensions in
terms of conditionability and arousability by empirical and correlations
methods, and animal and genetic models. Brand goes through some of the
inconsistencies in this grand project, in particular those pertaining to the P
dimension, then compares Cattell's "Big-six" approach to Eysenck's "Giant-
three" to illustrate approximate correspondences and discrepancies while
dismissing the McCrae and Costa "Big-five" model as it is "without any
psychological or physiological account whatsoever." An associated problem is
that high-IQ people seem to yield one factor solution and low-IQ people
another. Brand's approach to the problem of the number of dimensions is to
see whether high-IQ (Spearman's g) low-N people have a more differentiated
personality than low-IQ high-N people, that is, whether more dimensions are
needed to describe better-gifted stable individuals. He arrives tentatively at a
"Big-six" solution (involving g) and goes on to discuss their most likely basic
psychological underpinnings. Brand concludes on a pessimistic note when
putting Eysenckian psychology into perspective. He probes into questions such
as whether the search for psychological underpinnings will ever yield more
than previous mechanistic approaches; whether we now know for sure what is
basic to g', whether the Eysenckian attempt at reductionism will ever succeed
(see also chapter 25), and whether Eysenck could actually have had more luck
with his psychology. Perhaps, Brand ventures, the scientific bases of personality
are scattered among Cattellian "primaries," and perhaps the dimensional
variation needs explanation in terms of McDougallian links between individual
dynamics, purpose, and biological functions.
In chapter 3 Pickering, Corr, Powell, Kumari, Thornton, and Gray take up
the problem of individual differences in reactions to reinforcing stimuli. Like
Zuckerman, they first trace the influence of Pavlov on Eysenck with respect to
considering personality in terms of a small number of major dimensions, each
of which reflects individual differences on how subsystems of the brain
function. They then underline Eysenck's persistent insistence that personality
theory must be formulated in terms of the scientific method, that is, testable
hypotheses, empirical inquiry, and experimental variation. However, like
Brand in chapter 1, Pickering at al. see reason for pessimism: few have
followed Eysenck's lead and presented alternative testable accounts. All it
amounts to is a small change of names here, of location or interpretation of
factors there, but nothing that compares to Eysenck's massive underlying
structure of explanation. They then go on to discuss Gray's own reinforcement
sensitivity theory (RST) amendments to Eysenck's dimensional system.
Basically, Eysenckian dimensions are slightly rotated to conform more to the
lines of causal influence, stemming from separable subsystems of the brain.
These subsystems are believed to mediate different kinds of reactions to
reinforcing stimuli rather than different kinds of arousal, as Eysenck would
 General introduction xix

have it. The remainder of their chapter is devoted to an evaluation of how this
theory has fared in the light of a massive research program. Their main
empirical conclusion is that RST theory is more complex than first assumed.
Perhaps some equivocal test results may partly be ascribed to the absence of a
clear set of behavioral requirements for testing the full RST theory. They
therefore devote a section to discuss how to best test RST empirically. Another
reason for obtaining equivocal results is believed to stem from the application
of the same theoretical principle, but through the prism of different experi-
mental paradigms. The bewildering array of different outcomes is regretted by
the authors, in as much as it is indeed hard to doubt that the central nervous
system actually houses neural subsystems that influence individual differences
in sensitivity to reinforcement. The problem is pressing because Eysenck's own
biological theory for personality does not account for the personality/
reinforcement interactions that the Jeffrey Gray group concentrates on, and
which have also been observed by other researchers. The group sees no easy
solution in changing existing RST theory, and plans instead to develop a new
range of tasks based on sound theory to spot the underlying consistencies.
Strelau and Zawadzki discuss the three Eysenckian personality dimensions
in chapter 4 in terms of ancient and more contemporary dimensions of
temperament. They start out by pointing to Eysenck's inspiration from
Hippocrates, Galen, and the more recent Pavlov. They then proceed with an
examination of the nature of P, E, and N in order to see to which extent these
dimensions relates to temperamental dimensions. The authors note that P
seems to have a somewhat different status from E and N in this respect. Its
physiological, neurological, and biochemical status is as yet not as well defined
as is the basis for the other two dimensions. Eysenck tends to sees personality
as equivalent to temperament, but most temperament researchers finds that
temperament is only a part of personality, according to the authors. They rest
their case on studies of temperament, including some concentrating on (1) the
development of children's temperament, (2) Pavlovian properties of the
nervous system, (3) Rusalov's functional systems approach, (4) Zuckerman's
sensation-seeking theory, and (5) Strelau's own regulative theory of temp-
erament. The overall conclusion of the analysis is that almost all tempera-
mental scales correlate with either E, or N, or both, whereas P basically
correlates only with sensation seeking. They then settle for factor-analytic
examinations to see where the Eysenkian superfactors are located among other
temperamental characteristics. They find that E and N come out quite
independently of which temperamental scales are taken into account whereas
P seems not unequivocally to be of a temperamental nature. Strelau and
Zawadzki therefore imagine that P is a personality factor that differs from E
and N, and may, in fact, have much in common with one of Allport's
subdivisions of personality—character—and as such with a temperamental
disposition to inhibit impulses in accordance with a regulative principle.
xx The Scientific Study of Human Nature

David Nias takes up a controversial side of Eysenck's research, namely, the

psychology-medicine connection, in particular cancer-personality and
nutrition-intelligence relationships, but he also discusses smoking and illness
in chapter 5. A major question here is whether people experiencing difficulties
in expressing emotions run a higher risk than others for developing lung cancer
if exposed to cigarette smoke. Another question is whether psychological
intervention has an effect on the survival of cancer patients. Reflecting on
psychological factors in cancer, Nias first finds some support in the previous
literature for the assumption that cancer-prone personality traits exist and that
appropriate intervention may improve prognosis. He then goes on to discuss
Eysenck's collaboration with Grossarth-Maticek who previously had collected
a lot of data, and then in collaboration with Eysenck studied personality-
smoking-disease-cancer-death relationships. The positive results from these
studies came quickly under heavy criticism, and Eysenck took it upon himself
to respond. Nias suggests that interested readers consult a special issue of
Psychological Inquiry (Eysenck, 1991), in order to learn about the details of this
research. Having carefully reviewed the pros et contras Nias finds that the
harsh critique is not able to explain away the likelihood that there is a general
cancer-prone personality and that training with stress coping may prolong the
life of cancer patients and prevent the development of the disease in the first
place. Neither does he deny the need for sound replication studies. The latter
part of chapter 5 is devoted to a discussion of the research by Eysenck and
others on the effects of vitamin/mineral supplementation on intelligence in
children and of Grossarth-Maticek's investigation of whether vitamin/mineral
supplement prevents illness. Again Nias finds the observations promising but
stresses the need for better-controlled studies.
Psychoticism (P) is the third and latest major dimension in Eysenck's model
of personality. Sybil Eysenck has been very active in the development and
measurement of it. In chapter 6 she briefly covers the evolution of the P
dimension, and then goes on to discuss what is a proper test for P or, for that
sake, any personality dimension. She regrets that so many contemporary
models of personality remain content with factor analyses, because it makes
them basically subjective and untestable as they lack any objective criterion to
be judged by. After a presentation of the model for P and stressing its
dimensional rather than categorial nature and deep roots in biological factors,
Sybil Eysenck uses the rest of her chapter to discuss the application of the
model and to list the characteristic behavior of high P scorers. This takes her
through problems in the treatment of high P children, and the obstructions in
store for a psychotherapist dealing with the almost impossible-to-treat adult
personality-disordered high-P scorer: A successful reduction of N is met with
an increase in P, and vice versa. After an excursion into the impulsivity aspects
of the P dimension, Sybil Eysenck discusses the close relationship between
drug addition and high P score, and warns against recommending total
 General introduction xxi

abstinence. Fairly high P can be advantageous in some occupations and for

genius, but it also links to schizophrenia. A note of warning concludes the
chapter: Studies looking at any of the P, E, or N dimensions in isolation may
not get the true picture of personality, in particular if disregarding the lie score.
Crime has for a long time interested Hans Eysenck, dating at least back to
the early 1950s and leading up to his Crime and Personality book, published
first in 1964. His notion that criminals do not condition well had an immense
impact on many researchers, among them Adrian Raine. Raine first presents
Eysenck's conditioning theory of crime in chapter 7—that poor conditionability
means poor conscience and undersocialized, antisocial behavior. He then goes
on to examine the evidence for or against it, and finds great variation in
techniques, samples, and results. Raine stresses that Eysenck actually suggest-
ed a biosocial theory of crime, not just a genetic fancy. The basic idea is that the
environment-biology interaction is absolutely critical for the development of
crime. Raine goes on to present evidence in support of this view, and then
brings the theory to bear on brain imaging studies of prefrontal metabolic rates
in murderers with and without psychosocial dysfunction. Besides classical
conditions, underarousal also plays a role in criminal behavior according to
Eysenck, and this is confirmed in studies by Raine himself. He and his team
were able to confirm that high arousal, orienting, and conditionability may help
protect children from crime in adulthood. In the final section Raine takes on
Eysenck's shoes and wonders how a born-again Eysenck would tackle today's
growing problem of crime and violence in society if he had a new go in the
field.
Gisli Gudjonsson also deals with crime in chapter 8, but now with a focus on
how well Eysenck's theory explains the causes and cures of criminality.
Gudjonsson is particularly well equipped to tell this story because he is part
and parcel of it himself as co-author of The Causes and Cures of Criminality
(Eysenck & Gudjonsson, 1989). Historically, Eysenck never believed that
genes and personality alone could explain crime, and his most recent papers
even show increasing emphasis on the importance of social causes, so
Gudjonsson asks questions about the history of how precisely social factors
relate to psychological conditions disposing for crime. Well, extravert criminals
should not be punished but rather treated with a stimulant drug, according to
Eysenck in the 1960s. Later he emphasized that creating desirable conditioned
responses for prosocial behavior might actually work better than elimination of
undesirable conditioned responses. After reviewing some critique of Eysenck's
theory, Gudjonsson examines various empirical studies and finds that they
provide equivocal results. P is important for understanding crime, but it
sometimes fails to identify persistent and serious offenders and high P scorers
form only a small part of the criminal population. Gudjonsson concludes his
chapter with the question of whether Eysenck's theory has had more success in
stimulating research into the causes of crime than its cure.
xxii The Scientific Study of Human Nature

As must be pretty obvious by now, Hans Eysenck has never been afraid to
study sensitive matters. One of them is sex as it relates to personality. This
topic is dealt with in considerable detail by Glenn Wilson in chapter 9. Here
Wilson describes how Eysenck (in cooperation with himself), extended the
works of Kinsey and Masters and Johnson by investigating links between
constitutional personality factors, sexual attitudes, and behavior. The high P
scorer tends to practice early and impersonal or even hostile sex, the high E
scorer is promiscuous, and the high N scorer is highly driven but anxious and
ridden by dysfunctions. Wilson then goes on to describe personality correlates
to male-female differences, to the sexual behavior of criminal offenders,
including sex offenders, to sexual dysfunctions, and to marital satisfaction. The
general conclusion seems to be that personality, attitudes, and libido have
effects on marital happiness, so much that about 20% of the variance in marital
satisfaction may be explainable in terms of personality. What determines
personality? Wilson here draws upon three sources: genetic factors, sexual
conditioning, and sex hormones. Genetic factors appear to influence sex drive
and satisfaction but perhaps less for females than for males. The findings on
sexual conditioning are difficult to interpret though personality seems to be
involved, as are the effects of hormones. In conclusion, Wilson finds that
Eysenck's predictions from personality to sexual attitudes and behavior,
marital choices and satisfaction, and deviant behavior in general apply
reasonable well, and account for the findings much better than do rigid
moral prescriptions. Puritanism may suit the introvert and permissiveness the
extravert and high P scorer, and various kinds of personality mismatches may
predict marital unhappiness.
The last chapter in part I by William Revelle reviews the extensive literature
on impulsivity in relation to extraversion, personality in general, and
troublesome behavior in particular. This is not an easy task. Eysenck first
considered impulsivity part of extraversion, then of psychoticism. Others have
related it to sensation seeking and emotionality, or linked it to arousal and
cognitive performance with biological underpinnings. Gray rotated Eysenck's
E/N axes by 45° and found that impulsivity relates to high N, high E, and
represents a Behavioral Activating System, a solution Eysenck is not too happy
with. More problems were in store. Personality researchers have demonstrated
much creativity in developing new scales to measure old constructs, with the
result that there is considerable confusion as to whether different scales
measure identical constructs or different constructs are measured by scales
with similar labels. The unfortunate result is that the location of impulsivity in
various personality models varies by model specifications and theorists; add to
this that impulsivity most likely is a multidimensional construct, Revelle sighs.
All this causes problems for research on childhood precursors of adult impul-
ivity, delinquency, Attention Deficit-Hyperactivity disorders, relationships
among impulsivity-arousal-cognitive performance, stimulation preferences,
 General introduction xxiii

proneness to traffic violations and accidents, sensation seeking, conditioning,

and studies of level of performance in general, now requiring multiple levels of
analysis, as Revelle sees it. He concludes that impulsivity has not yet, despite its
importance for understanding individual differences in vital areas, found its
proper place in a multidimensional theory of personality.

Part II
Part II of the book is devoted to various aspects of research on intelligence,
such as its psychometric characteristics, it links to genes and biological
parameters, its geographical variation, its experimental and information
processing paradigms, and its degree of malleability by social engineering.
The list obviously is far from being representative for the field of intelligence,
and does not even pretend to cover the various topics Eysenck has been
actively involved in over the years.
Arthur Jensen first sketches the circumstances in the mid-1930s, when
Eysenck became familiar with the philosophy and quantitative technology of
the "London School." Here towering figures like Galton, Spearman, and Burt
strove to identify and objectively measure individual differences in mental and
behavioral traits. Eysenck immediately felt at home in this context, even if his
personal relationship to Burt could perhaps have been better. But soon he saw
the need for going beyond descriptive psychometrics and recommended an
experimental approach, first to test whether Gallon's notion of individual
differences in mental speed could explain individual differences in Spearman's
general ability, and then to see how electrical brain activity correlates with
ability. Reviewing the evidence, it is no wonder that Jensen is tempted to
rename the whole tradition: The Galton-Spearman-Eysenck school of
research on human mental ability.
In chapter 11 Jensen goes on to examine the psychometric characteristics of
intelligence, historically and more recently, and he certainly is at home here.
He has no problems with admitting that Binet and Simon pioneered the
construction of the first usable intelligence test, but he then calls attention to
the fact that their faulty "specificity doctrine" made them unable to understand
the essentials of their own test. Briefly, by counting the performance on each
specific test item one only sees the "noise;" test items typically do not correlate
well. The "signal"—or Spearman's g—can be distilled only by calculating item
covariance or true-score variance. This fact was not only unknown to Binet and
Simon but it is—lo and behold—also unknown to many contemporary
psychologists! The g factor is thus "invisible," and is neither particular
knowledge nor a specific characteristic of the test. It is a property of the brain,
according to Jensen, and has to ultimately be explained in terms of neuro-
physiological mechanisms. The way to get there is to accept the important
Hebbian-Eysenckian distinction between biological, psychometric, and
xxiv The Scientific Study of Human Nature

practical-social intelligence. The latter is too loosely defined to be of help in

revealing the causal nature of intelligence, so Jensen uses the rest of chapter 11
for a discussion of an important aspect of biological intelligence, namely,
reaction time and its relationship with psychometric g. According to Eysenck
(and Galton), the speed of information processing is the basis for individual
differences in g. Jensen and many others have by now spent more than 20 years
testing this hypothesis. The result is clear: speed of reaction and general
nonspeeded intelligence correlate from .30 to .50. As information processing
demands increase, the advantage of higher psychometric IQ becomes more
obvious. Clearly, speed and efficiency of information processing is at least part
of g, but why, asks Jensen. Is the reaction time measure a function of higher-
level cognitive processes, or does it reflect speed and efficiency of the neural
processes that causes differences in g? The answer is still blowing in the wind,
according to Jensen, but perhaps Spearman was right after all: "The final word
... must come from ... direct study of the human brain in its purely physical and
chemical aspects."
The question of a genetic basis for intelligence is sure to provoke negative
reactions in many quarters. But that cannot be helped. The question is
important, so anybody with a legitimate interest in the matter must be allowed
to ask, and to get answers, provided the usual stringent criteria of science are
conformed to. Philip A. Vernon does just this in chapter 12, where he discusses
the genetic and biological bases of intelligence. Vernon first compares various
measures of the heritability of intelligence, and then poses an interesting
question: do IQ and reaction time (RT) measures correlate because both
depend on the same genes and/or do environmental factors influence both?
The studies reviewed indicate a significant common genetic effect! He then
goes on to examine three biological correlates to intelligence—averaged
evoked potentials (AEP), cerebral glucose metabolic rates (CGMR), and
nerve conduction velocity (NCV). Various AEP measures correlate in general
with IQ but the connection is in need of a satisfactory theory. CGMR studies
suggest an interesting pattern: High-IQ subjects show higher brain activity at
rest than low-IQ subjects, but accomplish assigned cognitive tasks with a lower
consumption of energy. NCV studies give inconclusive results. Vernon notes
that phenotypic correlations between intelligence and personality are largely
attributable to common genetic factors, and then illustrates two new directions
in behavior genetics, namely group heritability and the molecular genetic
identification of quantitative trait loci.
Richard Lynn takes up still another controversial theme in chapter 13: the
geography of intelligence. Only individual and sex-related differences in
intelligence match this area in emotional stimulus value. From a scientific point
of view it really is saddening to see how often the ideology of equality for any
price makes many otherwise intelligent people literally blind to the rather
consequential differences in intelligence, that can easily be observed even by
 General introduction xxv

the untrained eye. Perhaps the greatest problem here is the assumption that
equality presumes biological identity, thereby confusing an apparently
benevolent moral and political agenda with the undeniable fact of biological
differences. In chapter 13 Lynn sets out to document that intelligence varies
systematically with geographical location. He then discusses the evidence for a
genetic basis for these differences and the theories for how they might have
come about during evolution as a function of differences in climate.
Lynn first points to a fact, unknown to many, that Eysenck originally held an
environmentalist view on the one standard deviation black-white IQ
difference, but later changed his opinion under the weight of several hundred
studies indicating that the differences might in part be genetically caused.
Lynn's own previous summaries of studies of the intelligence of the major races
suggested a particular ranking, with "Mongoloids" on the top, "Caucasoids" in
the middle, and "Negroids" at the bottom. The same ranking is found when
reaction time or brain size is measured. Eight recent studies confirm the
previous notion that Mongoloid populations in Asia have a median lead of
about 5 IQ points in front of the Caucasian median of 100. To that come
studies showing that American ethnic Mongoloids also have a 3-5-point lead.
Studies of the intelligence pattern indicate that Mongoloids have better spatial
than verbal intelligence. The median intelligence for seven recent studies of
sub-Saharan African Negroids is 69. This compares well with previous
estimates of a mean IQ of 70. Lynn asks whether race is a more important
determinant of intelligence than poverty or white rearing, and takes studies of
geographical and race differences in income as well as studies of transracial
adoption of black babies by white parents, to mean "yes." He concludes that
the worldwide consistency in available data provides clear evidence for a
genetic component in race differences in intelligence. To explain the
evolutionary basis for the genetic racial differentiation Lynn calls attention
to two theories: His own cold winters theory (Lynn, 1987; elaborated later by
himself and by Miller, e.g. 1991) and Rushton's (1991) climatic predictability
theory. Basically, cold climates put strong selection pressures on intelligence.
As Mongoloids experienced very cold winters in North East Asia, Caucasians
less cold winters in Europe, and Negroids a warm climate, the racial ranking in
intelligence makes sense in terms of difference in climate. In fact, there is a
third geo-climatic evolutionary model (Nyborg, 1987), which incorporates a
north-south gonadal hormone gradient (see also Nyborg, 1994, p. 25). Lynn
concludes the chapter by stressing that Eysenck was perfectly right when a
quarter of a century ago he found, on the basis of incomplete evidence, that the
black-white difference in intelligence has a genetic component. Subsequent
confluent research has only strengthened his original conclusion.
Ian Deary takes up the question of intelligence and information processing
in chapter 14. He first refers to Eysenck's unequivocal 1953 view that taxonomy
is all very well, but it will not bring us any closer to the mechanisms of
xxvi The Scientific Study of Human Nature

intelligence—be that speed of mental functioning or orderly sequences of brain

events. Deary then goes on to describe the history of attempts to go beyond the
standard psychometrics of intelligence and correlations, and to marry the
differential with the cognitive (information-processing) approach in a
desperate search for basic cognitive processes or processing primitives, only
to find obvious circularities. After this, Deary surveys the success of different
approaches to understand intelligence using information processing methods.
One technically rather demanding area is EEG and evoked potentials. These
measures do correlate with ability test scores, but they have revealed very little
about the brain/cognitive processes behind them. Research based on Stern-
berg's cognitive component approach led to poor intercomponent correlations,
correlations among presumed independent processes, and failure to prove the
existence of some said components or their life outside the IQ-type items
themselves. Steinberg's subtraction approach did not live up to its early
promises. Reaction time measures do correlate to a moderate extent with IQ
measures, but the methodology can be criticized on various grounds, and
Deary finds that the reaction time approach offers little hope with respect to
disentangling the sources of individual differences in intelligence. Inspection
time measures also correlate only to a moderate extent with intelligence, and
the theory and rationale for the inspection time index is undergoing wholesale
change at the moment. Overall, Deary finds only partial success for Eysenck's
long-standing insistence on the unification of the differential/correlational and
experimental/cognitive approaches to intelligence.
I invited Nathan Brody to write chapter 15 on the malleability and change in
intelligence. The book needed a balance, I thought, against what some might
find too heavy an emphasis on the biological side of intelligence. Much to my
relief Nat said "yes." He begins with Eysenck's moving account of how a black
autodidact—George Washington Carver born under sad conditions indeed—
nevertheless rose to the status of probably the greatest American biologist of
the last century. This history inspires Brody to take a careful look at the
conditions for change, continuity, and malleability of intelligence. He first
describes the Abecedarian Project, intended to increase intellectual perform-
ance in culturally disadvantaged children, mostly of African-American origin.
However, even this intense intellectual socialization experience could not
eradicate individual differences in child IQ, predictable from the mother's
intelligence. The environmental deprivation associated with deafness does not
seem to depress nonverbal IQ, though it, of course, affects verbal IQ. Chinese
and Japanese school children do better in tests for mathematics than American
and European children. However, despite a uniform exposure to a standard-
ized curriculum and a strong commitment to egalitarian and equal educational
exposure, Japanese children vary as much as American children exposed to
much more varied education. What about change with age? Brody here finds
that genes become increasingly relevant to intelligence over the adult life-span.
 General introduction xxvii

Having defined the Spearman general intelligence factor g, Brody is ready to

discuss various kinds of changes such as those dictated by the genotype and
changes predictable from infant measures. He then faces a key question: is
intelligence fixed once and for all? Well, later IQ can be predicted pretty well
from characteristics present prior to conception, in the absence of radical
environmental impacts, but there are naturally occurring changes in intel-
ligence with age and effects due to lack of proper schooling. The costly large-
scale attempts to raise IQ by formal intervention (e.g., project Headstart) fail
to show long-lasting effects, however, with the Abecedarian project as a
possible exception. Brody concludes that intelligence is malleable, but within
the limit of about one-third of a standard deviation, and then only in the case
of very intense and enduring preschool intervention. However, the question of
malleability is more complex than that. Children with high genotypic
intelligence exploit their environmental resources better than low-IQ child-
ren, and individuals tend to gravitate towards jobs commensurate with their
initial levels of ability. A curious by-product of the increasing egalitarism in
Western societies is that individual differences in genotypic intelligence will
increasingly determine the accomplishments of individuals. Summarizing the
evidence, Brody finds that it is our biological characteristics that individuate us
and that govern the ways in which we relate to our social worlds. This is what
Eysenck always believed, but it seems even more true today now we have good
evidence.

Part III
Part III takes up a tangled web of futher topics that has caught Eysenck's
interest. As Gibson says in his not entirely affectionate introduction: "Eysenck
is a man of many talents." It is therefore possible to provide the reader only
with fragments of his many endeavors and strides over half a century. Briefly,
one fragment is Irene Martin's chapter 16 on the role of classical condition in
personality theory. Another is an evaluation of what Eysenck has meant to our
understanding of psychopathology by Gordon Claridge in chapter 17, and the
psychophysical and psychophysiological underpinnings of extraversion and
arousal by Robert Stelmack in chapter 18. Eysenck has, during the past two
decennia, become increasingly interested in the roots of creativity and
geniality, culminating so far in his recent (1995) book on Genius. Besides
Sybil Eysenck (chapter 6), Philippe Rushton and Helmuth Nyborg treat
various aspects of this topic. Rushton does so in chapter 19 by painting a
(perhaps not particularly romantic) portrait of the genius, linking it to
psychoticism and intelligence, and (mis)conduct in university settings. Nyborg
presents a model for genius as an unstable molecular state in chapter 20.
Eysenck has not himself been particularly active in the area of occupational
psychology, but his personality theory has made an impact on the area. This
xxviii The Scientific Study of Human Nature

story is told in chapter 21 by Adrian Furnham. Chapter 22 touches upon a

highly controversial area nurtured by Eysenck. Here Suitbert Ertel reports on
Eysenck's involvement in the question of whether sunspot activity influences
creativity in a historical perspective. Much to the dismay of some of his more
orthodox friends, Eysenck has further been seriously involved in the study of
graphology, astrology, and parapsychology. His role in these areas is outlined
by the trio, David Nias, Geoffrey Dean, and Christoffer French in chapter 23.
Part III concludes with Arthur Jensen's precise portrait of Eysenck as teacher
and mentor. Personally I find that Art's account conies close to a declaration of
intellectual love between two towering figures in seach of uncompromising
objectivity in the foggy psychology of the twentieth century, heavily marked by
figures like Jung, Freud, and other uncensored metaphor makers.

Part IV
The epilog by Nyborg clearly strains the limit of a self-assumed editor. It
synapses on a number of highly idiosyncratic and selective impressions of what
psychology is, what Eysenck thinks it should be, and on what it could be in the
twenty-first century. I will not even try to excuse this self-indulgency, but will
resort to the Danish humorist, Robert Storm Petersen, who once said, "It is
difficult to predict, in particular about the future".

THE GRATEFULLY ACKNOWLEDGED

I am grateful to many for unfailing support and encouragement throughout

this project. First, thanks to the small group who enthusiastically backed up the
initiative from the beginning—Ian Deary, Irene Martin, Jan Strelau, Glenn
Wilson, and Marvin Zuckerman. Then, thanks to the many distinguished
contributors who, with such short notice, undertook the not too easy task of
putting Hans in proper perspective each from their vantage point. The speedy
publication was greatly facilitated through the professional assistance from
senior publishing editor Michele Wheaton and Marie Bowler from Elsevier
Science Ltd.
Let me conclude this introduction by stealing without remorse the last
phrase from Richard Lynn's preface to his 1981 Festschrift to Hans Eysenck.
Richard wrote:
It has been a privilege to edit this volume as a tribute to HJE. The Festschrift is
offered as some attempt to repay the inspiration and intellectual leadership he has
provided over more than a third of a century, not only to the contributors but to
numberless others working in the field of personality and other areas of psychology.
 General introduction xxix

The only amendment needed is to substitute "more than a third of a

century" with "more than half a century."
May your molecules continue to be with you, Hans!

HELMUTH NYBORG
University of Aarhus
Aarhus, Denmark

REFERENCES

Eysenck, H. J. (1964). Crime and personality (1st edn.). London: Methuen.

Eysenck, H. J. (1978). Personal communication. Cited in: H. B. Gibson. H. J. Eysenck's
contribution to hypnosis research. In Richard Lynn (Ed.), Dimensions of personality:
Papers in honour ofH. J. Eysenck (p. 228). Oxford: Pergamon.
Eysenck, H. J. (1991). Personality, stress, and disease: An interactionist perspective.
Psychological Inquiry, 2, 221-232.
Eysenck, H. J. & Gudjonsson, G. H. (1989). The causes and cures of criminality. New
York: Plenum.
Lynn, R. (1987). The intelligence of the Mongoloids: A psychometric, evolutionary and
neurological theory. Personality and Individual Differences, 8, 813-844.
Lynn, R. (1981). Dimensions of personality: Papers in honour of H. J. Eysenck. Oxford:
Pergamon.
Miller, E. M. (1991). Climate and intelligence. Mankind Quarterly, 32, 127-132.
Nyborg, H. (1987). Covariant trait development across races and within individuals:
Differential K theory, genes and hormones. Paper presented in the symposium on
"Biology-Genetics" at the Third Meeting of the International Society for the Study of
Individual Difference, Toronto, Canada, 18-22 June.
Nyborg, H. (1994). Hormones, sex and society: The science of physicology. Westport, CT:
Praeger.
Pearson, R. (Ed.) (1992). Shockley on eugenics and race: The application of science to the
solution of human problems. Washington, DC: Scott-Townsend.
Pearson, R. (1991). Race, intelligence and bias in Academe. Washington, DC: Scott-
Townsend.
Rushton, J. P. (1991). Do r/K strategies underlie human race differences? A reply to
Weizmann et al. Canadian Psychology, 32, 29-42.
Hans J. Eysenck at home, 14 June 1997.
PART I

Personality
This page intentionally left blank
Chapter 1
The psychobiological basis of personality

M. Zuckennan

We may say that personality is the central concept in our theory. The distal cause is
DNA, the proximal cause is the physiological, hormonal, neurological set of
intermediaries linking DNA to behaviour, and interacting with environmental
factors. (H. J. Eysenck, 1993)

1. HISTORICAL INTRODUCTION

Although little was known about the brain in relation to personality during the
first half of this century, and both psychoanalytic and learning theories
minimized the role of biological factors in personality, a few personality
theorists were convinced that personality traits were influenced by biological
trait variations. Murphy (1947) described his approach to personality as A
Biosocial Approach to Origins and Structure, but the biological factors are only
alluded to in a broad, abstract fashion and most of the book is concerned with
social, self, and psychoanalytic concepts. In his 1947 volume Eysenck devoted a
chapter to the relationship between physique (body-types) and personality, and
in his 1952 book there was a chapter on Heredity and environment containing a
twin study of the inheritance of neuroticism. However, most of the tests used to
define neuroticism and other factors were behavioral performance tests with
little theoretical rationale behind their selection and most were poorly
intercorrelated. The next year Eysenck (1953) applied the three-factor model
of personality to personality questionnaires, interest, and attitude tests with
more success and his approach began to shift to a theoretical one in which
questionnaires would be used to define personality factors in order to test
hypotheses deduced from theoretical interpretations of the factors and tested
in laboratory experiments.
Eysenck's 1957 book, The Dynamics of Anxiety and Hysteria, introduced
Hullian learning theory into the person-situation interaction with the
extraversion (E) dimension related to differences in reactive-inhibition and
neuroticism (N) to differences in drive. From these equations he was able to
4 Personality

predict differences in conditioning and other performance and perceptual

tasks. A chapter on Drugs and personality contrasts stimulant and depressant
drugs in their effects on introverts and extraverts as defined by their
characteristic behaviors.
Up to this time Eysenck's hypotheses about the relation between brain
physiology and personality were based on Pavlovian conceptions of individual
differences in excitatory and inhibitory balances in brain functioning. In his
1963 book, Experiments with Drugs he characterized the differences between
extraverts and introverts in an optimal level of stimulation, with extraverts
feeling best and functioning more efficiently at higher levels of stimulation, and
introverts feeling and functioning best at lower levels of stimulation. The
description of the interaction between intensity of stimulation and personality
resembled the contrast between the Pavlovian strong and weak nervous system
types, with the weak type (introverts) more sensitive to lower levels of
stimulation but more likely to show transmarginal inhibition at high intensities
of stimulation where the strong nervous system types (extraverts) could
continue to respond (see chapter 4)
The theoretical paradigm reached full maturity in The Biological Basis of
Personality (Eysenck, 1967). It incorporated advances made in the neuro-
sciences regarding the role of brain stem and limbic systems in cortical arousal
and emotional reactions. The optimal level of stimulation was related to an
optimal level of arousal (Hebb, 1955) based on the functioning of the
reticulocortical activating system (Moruzzi & Magoun, 1949). This system
functioned as a homeostat regulating the stimulation input to the cortex in
terms of current levels of arousal (Lindsley, 1957). The set point at which the
homeostat shut down input (transmarginal inhibition) was suggested to differ
in introverts and extraverts, being lower in the former and higher in the latter.
Previously, Eysenck had attributed differences in N, or emotionality, to
differences in reactivity of the peripheral autonomic nervous system. In 1967
he recognized the role of the visceral (limbic system) brain in initiating
autonomic and hormonal system changes and producing a type of subcortical
arousal associated with emotions. Individual differences in neuroticism were
now attributed to differential thresholds in the "visceral" brain with interaction
between the two systems (visceral brain arousal potentiating reticulocortical
arousal). Little differentiation was made among the various structures and
pathways in the limbic brain.
The third dimension of personality, psychoticism (P), received only the
briefest mention in the 1967 volume. Eysenck suggested that there is a broad
genetic factor of psychotic predisposition with separate genes determining the
precise form (schizophrenic or manic-depressive) of the psychosis. A more
extensive review of twin studies of E and N, as measured by tests, suggested
that about 50% of individual differences in these traits is determined by
heredity.
 The psychobiological basis of personality 5

Research on P as a dimension of personality, initiated in studies by Eysenck

(1956), had not advanced far because the standard Eysenck questionnaires
only included scales for E and N. In the 1970s a scale for P was published
(Eysenck, S. B. G. & Eysenck, H. J., 1972) and a fuller exposition of the theory
and findings on the P dimension was published later (H. J. Eysenck & S. B. G.
Eysenck, 1976). The dimension is now conceived of as a broad, additive-
polygenetic trait underlying all psychotic disorders including mood and
schizophrenic types (see chapter 6). The trait may be linked to creativity in
normals, thereby accounting for its survival despite the low reproductive rate of
schizophrenics. (Critics have noted anomalies in the results on the scale in
relation to its title and concept; for instance, the fact that prisoners, drug
addicts, alcoholics, and personality disorders score higher than psychotics on P,
suggesting that psychopathy or socialization may have been a more appropriate
name for the dimension than psychoticism (Bishop, 1977; Block, 1977;
Zuckerman, 1989).)
A second major exposition of the mature theory in a book by H. J. Eysenck
and M. W. Eysenck (1985) shows little change in the biological basis of
personality and the arousal theory of E and N. The main addition is the
incorporation of the P dimension and some redistribution of narrower traits in
the hierarchical system. Recent statements of the theory (e.g., H. J. Eysenck,
1991,1993) have recognized new biochemical and psychophysiological findings
but have not incorporated them into the system or considered modifications of
the arousal theory (Eysenck, 1987).

2. EYSENCK'S INFLUENCES

My early work in personality had been largely in assessment and although I

taught personality courses my theoretical interests were not strongly biological.
I had worked in interdisciplinary research settings twice (1956-1960 and 1963-
1969) with biochemists and endocrinologists, and studied the psychophysiology
of emotions, but I largely considered these kinds of measures as useful indica-
tors of internal emotional states rather than as trait markers for fundamental
dimensions of personality. From 1958 to 1969 I experimented in the area of
sensory deprivation. In the early 1960s I felt the need for a new personality
instrument that might assess the need for stimulation that made even short-
term sensory deprivation stressful for many subjects. Hebb's (1955) concept of
an optimal level of arousal and Berlyne's concept of the arousal potential of
stimuli as a function of their novelty, intensity, and complexity provided a
conceptual basis for items written for the first form of the Sensation Seeking
Scale (SSS, Zuckerman, Kolin, Price, & Zoob, 1964). In the mid-1960s I was
asked to write a theoretical chapter for a book edited by John Zubek (1969) and
began a more intensive consideration of arousal and its brain sources. Lindsley
6 Personality

(1961) had postulated that differences in reticulocortical activating system

sensitivity might be a source of differences in response to sensory deprivation
and I made this part of my first biological model for the trait.
In 1966 I stopped in London on my way to an international psychology
conference in Moscow. This was my first meeting with Hans. I was eager to
share with him my idea of basing a trait of sensation seeking on optimal levels of
stimulation and arousal. I was somewhat consternated when he informed me
that he was making the optimal level idea the central construct in his theory of
introversion-extraversion. I had not read his 1963 book (Eysenck, 1963) on
Experiments with Drugs in which he first expounded the idea. In discussing the
sensation-seeking trait with him, it became clear that he regarded it as a subtrait
of E. I went along with this idea for a time despite the fact that eight studies
conducted in the 1970s yielded a median correlation of only .25 between E and
the SSS (Zuckerman, 1979). H. J. Eysenck and S. B. G. Eysenck (1976) decided
that sensation seeking was a subtrait of impulsivity which was in turn a subtrait
of E. Buss and Plomin (1975) considered sensation seeking a subtrait of
impulsrvity, but impulsivity was one of their four basic temperaments and
separate from sociability (extraversion). Later they decided to drop impulsivity
as a basic temperament, leaving only three: emotionality, activity, and sociability
(Buss & Plomin, 1984).

2.1 Personality structure

In the 1980s I began a series of factor analytic studies to determine just where
sensation seeking did lie among the basic dimensions of personality. Using a
variety of personality scales including the four subscales of the SSS, we
discovered that impulsivity and sensation-seeking scales together formed one
of the three major factors of personality, defined at the negative pole by
socialization, responsibility, and need for cognitive structure (Zuckerman,
Kuhlman, & Camac, 1988). The P scale fell right on the axis defining this factor
and was one of the highest loading scales at the positive end of the dimension.
We named the factor P-Impulsive Unsocialized Sensation Seeking (P-
ImpUSS). The P in the label suggests that impulsivity, lack of socialization,
and sensation seeking form the personality core of the P dimension. The factor
has been replicated in subsequent studies factoring both scales and items
(Zuckerman, Kuhlman, Thornquist, & Kiers, 1991; Zuckerman, Kuhlman,
Joireman, Teta, & Kraft, 1993). On the basis of these results showing such
close identification of P with impulsivity, socialization, and sensation seeking
and other findings on the genetic bases of the psychoses, I suggested that P was
perhaps an inaccurate construct for the dimension and that psychopathy or
socialization might be a better one (Zuckerman, 1989).
 The psychobiological basis of personality 7

On the bases of these factor analyses, we have constructed a new test to

assess five reliable factors: sociability, neuroticism/anxiety, impulsive sensation
seeking, aggression/hostility, and activity (Zuckerman et al., 1993). The first
two factors correlate very highly with Eysenck's E and N. ImpSS correlates
highly with P, with only some secondary correlation with E. In a three-factor
solution together with NEO scales (Costa & McCrae, 1985), Eysenck's E, N,
and P constitute the best markers for the three basic three dimensions cutting
across all scales. But we feel that the separation of aggression/hostility (the
obverse of agreeableness in the NEO) from the broad N factor is important
from a biological and ethological viewpoint. The biological and behavioral
mechanisms for anxiety and aggression are quite different. Activity is a
component of extraversion in a three-factor model, but constitutes a basic
dimension in Buss and Plomin's (1984) temperament classification and is a
basic trait in most analyses of child temperament and a basic variable hi most
comparative studies of other species. Our five-factor model (called the
"alternative-5" to distinguish it from the "big-5") may prove to be more useful
than Eysenck's big-3, or Costa and McCrae's (1985) big-5 in studies of the
biological bases of temperament or basic personality.

2.2 Genetics
In 1975 I spent my sabbatical with Hans and Sybil Eysenck at the Institute of
Psychiatry in London. We decided to undertake three projects involving
sensation seeking. The first was to develop a new and shorter form for the SSS
with a balanced total score based on the sum of scales representing each of the
four factors discovered in previous factor analyses and to compare English and
American populations on the scales. The second was to examine the relation-
ships between the SSS and the EPQ scales (E, N, and P). The third was to
assess the heritability of sensation seeking using a large twin sample already
tested on the EPQ. The SSS form V, now the most widely used one, was
developed from item factor analyses in both countries (Zuckerman, S. B. G.
Eysenck, & H. J. Eysenck, 1978). Males in America, England, and Scotland did
not differ on SSS Total scores, but English women scored lower than American
and Scottish women. Men scored higher than women in all three countries, a
sex difference now widely replicated in many other countries.
Our most interesting results came from the twin study (Fulker, Eysenck, &
Zuckerman, 1980). Analyses of the twin data using the Jinks and Fulker's
(1970) biometric method showed that 58% of the overall trait could be
accounted for by genetic factors and the rest was due to nonshared environ-
ment and errors of measurement. Lykken (personal communication, 1995)
examined the correlations of SSS total scores in twins who were separated at,
or shortly after, birth and adopted into different families where they spent all
of their formative years. The correlation of the SSS for separated identical
8 Personality

twins was .54 compared to the correlation obtained by Fulker et al. (1980) of
.58. Since the correlation between separated identical twins is a direct measure
of heritability, this would mean that the heritability was 54%. The separated
fraternal twins correlation of .34 must be doubled to yield the heritability so
that it would be 68%. If we average these two estimates they come to 59%,
almost exactly the same as the estimate from twins raised together. The range
of heritability for most personality traits is 30-60% so that the heritability of
sensation seeking is at the limits of the high range.
The lack of difference between the correlations for twins raised apart in
different adoptive homes and those raised together by their biological parents
indicates the lack of influence of a shared environment. This is now a common
finding for many personality traits and challenges the theories which ascribe
major importance to an early shared family environment in the shaping of
personality. We have found no correlations between their parental treatment,
as reported by late adolescents, and the impulsive sensation-seeking trait,
although there were significant correlations with neuroticism and sociability
(Kraft, 1995).
Just as the findings of biometric behavioral genetics were becoming
repetitive and even boring, the methodological breakthroughs in molecular
genetics have allowed us to ask a more interesting question than "what are the
relative influences of heredity and environment?" The new question is "which
major genes are associated with personality traits?" (Plomin, 1995). Until quite
recently the methods of molecular genetics have been applied to medical and
psychiatric disorders, with hardly any replicable results for the latter. However,
two independent studies have identified a relationship between measures of
novelty seeking and different forms of the dopamine receptor gene, D4DR
(Benjamin et al., 1996; Ebstein et al., 1996). [Novelty seeking, as measured by
Cloninger's questionnaire, correlates .68 with Impulsive Sensation Seeking,
indicating near equivalence of the two scales (Zuckerman & Cloninger, 1996).]
However, variations of populations and methods may result in inconsistent
findings even when working with genes. Two failures of replication were
reported at a recent psychiatric genetics meeting which I attended. Apart from
the differences in populations and methods, the D4DR gene accounts for only
10% of the genetic variance in novelty seeking so that there are other major
genes remaining to be discovered. The particular combination of genes
influential in the trait would be expected to vary between populations. The
theory and findings on the psychopharmacology of sensation seeking suggest
where to look next for these genes.

2.3 Arousal
Eysenck and I had both been influenced by the constructs of the "optimal level
of stimulation" and "optimal level of arousal" in our theories of extraversion
 The psychobiological basis of personality 9

and sensation seeking, although Eysenck emphasized mainly intensity of

stimulation, whereas following Berlyne, I suggested that the stimulus qualities
of novelty, complexity, and the need for change or variation as well as intensity
were motivating qualities for sensation seekers. The first hypothesis for
sensation seeking, as for E, was that sensation seekers sought intense and/or
novel stimulation because they were generally underaroused relative to their
higher optimal level of arousal. Psychophysiological methods, like skin con-
ductance, heart-rate, and EEG measures were first used to assess basal arousal
and arousability in response to stimuli (Zuckerman, 1990). We found no
differences in basal level of arousal however measured. However, high
sensation seekers were more aroused (skin conductance response) by novel
stimuli, but when the stimuli were repeated they showed no more arousal than
low sensation seekers. This was not a difference in gradual habituation to the
stimulus, but a rather abrupt reduction in arousal occurring the second time
the stimulus was presented. Somewhat more persistent reactions were found
when the stimuli had a content of intrinsic interest to high sensation seekers
(Smith, Perlstein, Davidson, & Michael, 1986).
Smith (1983) and Stelmack (1990), reviewing the literature on E, reported
that introverts and extraverts do not differ on general arousal in normal basal
conditions, but the measures of skin conductance indicated that introverts tend
to be more arousable than extraverts to stimuli of moderate intensity.
Heart rate (HR) measures varied as a function of the intensity of the stimuli
(auditory). Low-intensity stimuli elicit HR deceleration in most subjects, and
high-intensity stimuli tend to produce HR acceleration. Stimuli of intermediate
intensities produce more variable responses: High sensation seekers tend to
show HR deceleration and low sensation seekers show HR acceleration.
Deceleration is associated with the orienting response or attention to the
stimulus and acceleration may represent either a startle or a defensive res-
ponse. However, these differences, like the SCR differences related to
sensation seeking, depend on the novelty of the stimuli; they are only apparent
on the first or second trials in a sequence of stimulus repetitions.
Cortical response measures are most relevant to the arousal or arousability
hypotheses because these were initially based on arousal in the reticulocortical
system. No differences related to sensation seeking have been reported in
arousal as measured by EEG spectrum analyses during basal conditions. The
results for E are mixed and inconclusive (Gale & Edwards, 1986; O'Gorman,
1984). However, the cortical evoked potential (EP) has been shown to vary
with stimulus intensity in a different way for high and low sensation seekers
(Zuckerman, 1990). The high sensation seekers (particularly on the Dis-
inhibition subscale) show a strongly increasing EP amplitude evoked by visual
or auditory stimuli, whereas the low sensation seekers show little EP increase
as a function of stimulus intensity, and at the highest intensities they sometimes
show a sharp reduction in EP response. The former pattern has been called
10 Personality

"augmenting" and the latter has been described as "reducing," but the relative
reactions are continuously and normally distributed ranging from extreme
augmenting to extreme reducing (Buchsbaum, 1971).
Augmenting as a physiological trait has been associated with certain types of
psychopathology also characterized by high sensation seeking: bipolar disorder,
alcoholism, delinquency, and criminality (Zuckerman, 1994). Animal models for
augmenting and reducing have shown mat the psychophysiological trait is assoc-
iated with behavioral traits analogous to sensation seeking. Augmenter cats tend
to be more exploratory, active, aggressive, and approach novel stimuli (Lukas &
Siegel, 1977; Saxton, Siegel, & Lukas, 1987). In an experimental paradigm,
augmenting cats are less able to inhibit behavior in a paradigm that punishes a
too rapid rate of responding by withholding reinforcement (Saxton et al., 1987).
A strain of rats characterized by EP augmenting showed more exploration and
less emotionality in the open-field test than another strain characterized by
reducing. The augmenting strain of rats was generally less fearful, a greater
tendency to drink water laced with alcohol, and the females were less nurturing to
their pups. Most interesting is the fact that the augmenting type rats were more
responsive to high-intensity stimulation of the "reward centers" of the brain.
These centers depend on the neurotransmitter dopamine being activated and
they are the sites of action of drugs abused by high sensation seekers.

2.4 Psychopharmacology
My interest in psychopharmacology began with the findings of a negative
relationship between the enzyme monoamine oxidase (MAO) and sensation
seeking (Murphy et al., 1977; Schooler, Zahn, Murphy, & Buchsbaum., 1978).
This has been a fairly well-replicated result (significant in nine of 13 studies)
although the typical magnitude of the correlation (-.24) is not high
(Zuckerman, 1994). The highest and most consistent findings have been with
the SSS subscale, Disinhibition, the one showing the strongest relationship
with EP augmenting. In fact, augmenters tend to have lower MAO levels than
reducers (von Knorring & Penis, 1981). The MAO type B, assayed from blood
platelets, has been found to have a particular affinity for dopamine. Its action is
catabolic and its function seems to be to regulate the level of the neuro-
transmitter stored in the neuron. Low MAO levels in humans have been found
in those with histories of criminality, drug, alcohol, and tobacco use and abuse,
antisocial and borderline personality disorders, and bipolar clinical disorder.
All of these disorders have been associated with high levels of sensation
seeking. Low MAO levels in monkeys have been associated with social
dominance, sexual, and play activity.
Interest in the role of the monoamines in personality led me to the work of
Gray (1982, 1987) and Stein (1978), both of whom worked primarily with rats.
Both of these investigators ascribed a central role to brain dopamine in
 The psychobiological basis of personality 11

approach motivation, but Gray also regarded dopamine as essential in the

reward mechanisms of the mesolimbic system. I evaluated the comparative
literature on the monoamines in my first book on sensation seeking
(Zuckerman, 1979) and in my 1984 paper in Behavioral and Brain Sciences
(Zuckerman, 1984) and entertained the idea of a new optimal level theory, but
one based on the brain catecholamine systems (dopaminergic and noradren-
ergic) not the reticulocortical system. At optimal levels of catecholamine
activity (CSA), mood is positive, and exploratory activity and social interaction
are high. Psychopathology appears at extreme levels of CSA activity,
depression at the low end and panic, paranoia, and aggression at the high
end of the range of CSA. More recently I have formulated a model (Figure 1.1)
which suggests that the three monoamine neurotransmitters underlie three
behavioral mechanisms involved in four of the major personality traits,
dopamine the approach, serotonin the behavioral inhibition, and norepin-
ephrine (and GABA) the arousal mechanisms (Zuckerman, 1991, 1995). The
enzymes, such as MAO, and hormones, such as testosterone and cortisol, also
play a role in personality traits, either through their regulatory effects on
neurotransmitters or direct effects on brain and behavioral mechanisms.
Although Gray has been more of an influence than Eysenck in my
theoretical shift from cortical psychophysiology to neurochemical brain
pathways, it was Eysenck who originally suggested the value of a comparative
approach to personality, as in his studies of Maudsley emotionally reactive and
unreactive rats as a model for neuroticism and a study of primary dimensions

E-Sociability P-ImpUSS N-Anxiety

i VApproach ^
*+
' t-
^ Inhibition Arousal

'/-/ '.*v ' t*. i

•Norepinephrine
VGABA
Gonadal
TT
Hormones _
^ nfLf\\J
_, _,
TypeB
DB
\-
Endorphins
Figure 1.1. A psychopharmacological model for extraversion, impulsive unsocialized sensation
seeking (ImpUSS) and neuroticism-anxiety (N-Anxiety) with underlying behavioral (Approach,
Inhibition, and Arousal) and physiological mechanisms and neurotransmitters, enzymes, and
hormones involved. MAO = monoamine oxidase, DBH = dopamine-beta-hydroxylase, GABA =
gamma-aminobutyric acid. Interactions between behavioral and biochemical factors indicated (+)
for agonistic and (—) for antagonistic actions. Single-headed arrows indicate the hypothesized
direction of influence; double-headed arrows indicate a two-way interaction between the factors.
12 Personality

of behavior in monkeys (Chamove, Eysenck, & Harlow, 1972). Besides Gray, I

attribute my interest in cross-species, biologically based motives like approach
and withdrawal and seeking and avoidance to Schneirla (1959), a teacher from
my graduate school days. Berlyne (1967, 1971) was also an early influence in
the idea that the relative activity of brain reward and aversion systems
determines the arousal potential and its optimal level. My model broadened
Schneirla's seeking motivation as a function of stimulus intensity to include
novelty as a provoker of approach, with variation in optimal levels of arousal
(catecholamines) as the basis for impulsive sensation seeking.

3. THE FUTURE

The future of the psychobiology of personality is already discernible in the new

methods now in their infancy. In the not too distant future we will be able to
identify some major genes associated with the basic personality traits in
humans and behavioral traits in animals. I predict we will find considerable
overlap in the genes associated with traits like impulsive sensation seeking and
sociability, but somewhat more specificity in the genes associated with specific
behavioral mechanisms. After all it was the behavioral mechanisms, not
personality traits, which were selected in the long evolutionary history of
hominids and other species preceding them. Discovery of these genes and an
understanding of what biological structures and functions they shape will
enable us to fill in the many intermediate levels between genotype and
phenotype. Another area which must develop is that of genotype-environment
correlations. Simple interaction models do not work. How do individuals select
activities, vocations, peers, mates, etc. on the basis of their own genetically
influenced traits? The answers to these questions will require more imaginative
research designs than we have used thus far.
Most of our neurochemical theories have been based on differences in
activity in specific pathways mediated by neurotransmitters. But interests in
molecular biology have shifted to the various receptors mediating the activity
of neurotransmitters systems. New variants of receptors are discovered for
each neurotransmitter and the situation has become even more complex.
However, there are functional and structural bases for this variation in
receptors and when we understand these we will better understand the
variations in the behavior mechanisms and traits related to them. Interactions
between neurotransmitters must be studied. The ancient Greeks had it right
when they said it was the balance between the humors that determined
temperaments and mental disorders. Most studies have looked at only one
neurotransmitter or hormone at a time even though it is clear that more than
 The psychobiological basis of personality 13

one is involved in disorders like depression. We must overcome the type of

phrenological thinking that insists there is only one "humor" associated with
one personality trait.
The advances in neuroimaging, especially neurotransmitter-specific imaging,
will enable us to test the physiological reactions postulated in the human brain.
At present the few personality studies have been incidental to studies of
psychiatric disorders. The techniques have been too expensive to devote to
studies of the large samples needed to understand personality as a function in
brain physiology. As the cost comes down and more laboratories are able to
participate, and as the resolution of brain images becomes sharper, we can
anticipate direct testing of the neurochemical hypotheses now proposed for
personality traits. For instance, we could see differential dopamine activity of
high and low impulsive sensation seekers in the medial forebrain bundle when
certain kinds of positively arousing stimuli are presented.
Eysenck's paradigm has accomplished several things in the field of
personality psychology. First, it recognized the importance of factor analysis
combined with theory in the construction of tests to be used as instruments of
theory. It stands in contrast to recent attempts to define basic dimensions of
personality in totally atheoretical terms. Second, it promoted the biometric
study of the genetic and environmental bases of personality. The results of
these studies have challenged the social-environmental schools which regarded
personality as entirely a product of the family and social environments. Third,
it emphasized the role of biochemical and physiological variation in personality
and the brain systems which are involved in these differences. Finally, it was
the first theory to present a complete account of personality from the basic
foundation in the genome to the complex social interactions and behavioral
traits which characterize individuals.

REFERENCES

Benjamin, J., Li. L., Patterson, G, Greenberg, B. D., Murphy, D. L., & Hammer, D. H.
(1996). Population and familial association between the D4 dopamine receptor gene
and measures of sensation seeking. Nature Genetics, 12, 81-84.
Berlyne, D. E. (1960). Conflict, arousal and curiosity. New York: McGraw-Hill.
Berlyne, D. E. (1967). Arousal and reinforcement. In D. Levine (Ed.), Nebraska
symposium on motivation (Vol. 15). Lincoln, NE: University of Nebraska Press.
Berlyne, D. E. (1971). Aesthetics and psychobiology. Englewood Cliffs, NJ: Prentice-
Hall.
Bishop, D. V. M. (1977). The P scale and psychosis. Journal of Abnormal Psychology, 86,
127-134.
Block, J. (1977). The Eysencks and psychoticism. Journal of Abnormal Psychology, 86,
653-654.
Buchsbaum, M. S. (1971). Neural events and the psychophysical law. Science, 172, 502.
14 Personality

Buss, A. H., & Plomin, R. (1975). A temperament theory of personality development. New
York: Wiley Interscience.
Buss, A. H., & Plomin, R. (1984). Temperament: Early developing traits of personality.
Hillsdale, NJ: Erlbaum.
Chamove, A. S., Eysenck, H. J., & Harlow, H. F. (1972). Factor analyses of Rhesus
social behaviour. Quarterly Journal of Experimental Psychology, 24, 496-504.
Costa, P. T. & McCrae, R. R. (1985). The NEO personality inventory manual. Odessa,
FL: Psychological Assessment Resources.
Ebstein, R. P., Novick, O., Umansky, R., Priel, B., Osher, Y., Blaine, D., Bennett, E. R.,
Nemanov, L., Katz, M., & Belmaker, R. H. (1996). Dopamine D4 receptor (D4DR)
exon HI polymorphism associated with the human personality trait of novelty
seeking. Nature Genetics, 12, 78-80.
Eysenck, H. J. (1947). Dimensions of personality. New York: Praeger.
Eysenck, H. J. (1952). The scientific study of personality. New York: Praeger.
Eysenck, H. J. (1953). The structure of human personality. New York: Wiley.
Eysenck, H. J. (1957). The dynamics of anxiety and hysteria. New York: Praeger.
Eysenck, H. J. (1963). Experiments with drugs. New York: Pergamon.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Charles C.
Thomas.
Eysenck, H. J. (1987). Arousal and personality: The origins of a theory. In J. Strelau &
H. J. Eysenck (Eds.), Personality dimensions and arousal (pp. 1-13). New York:
Plenum.
Eysenck, H. J. (1991). Dimensions of personality: The biosocial approach to personality.
In J. Strelau & A. Angleitner (Eds.), Explorations in temperament: International
perspectives in theory and measurement (pp. 87-103). London: Plenum.
Eysenck, H. J. (1993). From DNA to social behaviour: Conditions for a paradigm of
personality research. In J. Hettema & I. J. Deary (Eds.), Foundations of personality
(pp. 55-66). Dordrecht: Kluwer Academic.
Eysenck, H. J. & Eysenck, M. W. (1985). Personality and individual differences: A natural
science approach. New York: Plenum.
Eysenck, H. J., & Eysenck, S. B. G. (1976). Psychoticism as a dimension of personality.
New York: Crane, Russak, & Co.
Eysenck, S. B. G. (1956). Neurosis and psychosis. An experimental analysis. Journal of
Mental Science, 102, 517-529.
Eysenck, S. B. G., & Eysenck, H. J. (1972). The questionnaire measurement of
psychoticism. Psychological Medicine, 1, 50-55.
Fulker, D. W., Eysenck, S. B. G., & Zuckerman, M. (1980). The genetics of sensation
seeking. Journal of Personality Research, 14, 261-281.
Gale, A., & Edwards, J. A. (1986). Individual differences. In M. G. H. Coles, E.
Donchin, & S. W. Forges (Eds.), Psychophysiology: Systems, processes and applications
(pp. 431-507). New York: Guilford.
Gray, J. A. (1982). The neuropsychology of anxiety: An enquiry into the functions of the
septohippocampal system. New York: Oxford University Press.
Gray, J. A. (1987). The neuropsychology of emotions and personality. In S. M. Stahl, S.
D. Iverson, & E. C. Goodman (Eds.), Cognitive neurochemistry (pp. 171-190).
Oxford: Oxford University Press.
 The psychobiological basis of personality 15

Hebb, D. O. (1955). Drives and the C.N.S. (conceptual nervous system). Psychological
Review, 62, 243-254.
Jinks, J. L., & Fulker, D. W. (1970). Comparison of the biometrical genetical, MAVA,
and the classical approaches to the analysis of human behavior. Psychological Bulletin,
73, 311-349.
Kraft, M. R. Jr. (1995). Perceived parental behavior of biological parents and stepparents
reported by young adults, and relationships between perceived parenting and personality.
Unpublished master's thesis, University of Delaware, 1995.
Lindsley, D. B. (1957). Psychphysiology and motivation. In M. R. Jones (Ed.), Nebraska
Symposium on Motivation: Vol. 5. Lincoln, NE: University of Nebraska Press.
Lindsley, D. B. (1961). Common factors in sensory deprivation, sensory distortion and
sensory overload. In P. Solomon, P. E. Kubsansky, P. H. Leiderman, J. H.
Mendelson, R. Trumbull, & D. Wexler (Eds.), Sensory deprivation (pp. 174-194).
Cambridge, MA: Harvard University Press.
Lukas, J. H. & Siegel, J. (1977). Cortical mechanisms that augment or reduce evoked
potentials in cats. Science, 196, 73-75.
Moruzzi, G. & Magoun, H. W. (1949). Brain stem reticular formation and activation of
the EEG. EEC Clinical Neurophysiology, 1, 455-473.
Murphy, D. L., Belmaker, R. H., Buchsbaum, M. S., Martin, N. F., Ciaranello, K., &
Wyatt, R. J. (1977). Biogenic amine related enzymes and personality variations in
normals. Psychological Medicine, 7, 149-157.
Murphy, G. (1947). Personality: A biosocial approach to origins and structure. New York:
Harper.
O'Gorman, J. G. (1984). Extraversion and the EEG. I: An evaluation of Gale's
hypothesis. Biological Psychology, 19, 95-112.
Plomin, R. (1995). Molecular genetics and psychology. Current Directions in
Psychological Science, 4, 114—117.
Saxton, P. M., Siegel, J., Lukas, J. H. (1987). Visual evoked potential augmenting-
reducing slopes in cats: 2. Correlations with behavior. Personality and Individual
Differences, 8, 511-519
Schneirla, T. C. (1959). An evolutionary and developmental theory of biphasic
processes underlying approach and withdrawal. In M. J. Jones (Ed.), Nebraska
Symposium on Motivation: Vol. 7 (pp. 1-42). Lincoln, NE: University of Nebraska
Press.
Schooler, C., Zahn, T. P., Murphy, D. L., & Buchsbaum, M. S. (1978). Psychological
correlates of monoamine oxidase in normals. Journal of Nervous and Mental Disease,
166, 177-186.
Smith, B. D. (1983). Extraversion and electrodermal activity: Arousability and the
inverted-U. Personality and Individual Differences, 4, 411—419.
Smith, B. D., Perlstein, W. M., Davidson, R. A., & Michael, K. (1986). Sensation
seeking: Differential effects of relevant novel stimulation on electrodermal activity.
Personality and Individual Differences, 4, 445-452.
Stein, L. (1978). Reward transmitters: Catecholamines and opioid peptides. In M. A.
Lipton, A. DiMascio, & K. F. Killam (Eds.), Psychopharmacology: A generation of
progress (pp. 569-581). New York: Raven Press.
Stelmack, R. M. (1990). Biological bases of extraversion: Psychophysiological evidence.
Journal of Personality, 58, 293-311.
16 Personality

von Knorring, L., & Ferris, C. (1981). Biochemistry of the augmenting-reducing

response in visual evoked potentials. Neuropsychobiology, 7, 1-8.
Zubek, J. (Ed.) (1969). Sensory deprivation: Fifteen years of research. New York:
Appleton-Century.
Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal Hillsdale,
NJ: Erlbaum.
Zuckerman, M. (1984). Sensation seeking: A comparative approach to a human trait.
Behavioral and Brain Sciences, 7, 413-471.
Zuckerman, M. (1989). Personality in the third dimension: A psychobiological
approach. Personality and Individual Differences, 10, 391-418.
Zuckennan, M. (1990). The psychophysiology of sensation seeking. Journal of
Personality, 58, 313-345.
Zuckerman, M. (1991). Psychobiology of personality. Cambridge, U.K.: Cambridge
University Press.
Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking.
New York: Cambridge University Press.
Zuckerman, M. (1995). Good and bad humors: Biochemical bases of personality and its
disorders. Psychological Science, 6, 325-332.
Zuckerman, M., & Cloninger, C. R. (1996). Relationships between Cloninger's,
Zuckerman's, and Eysenck's dimensions of personality. Personality and Invidividual
Differences, 21, 283-285.
Zuckerman, M., Eysenck, S. B. G., & Eysenck, H. J. (1978). Sensation seeking in
England and America: Cross-cultural, age, and sex comparisons. Journal of
Consulting and Clinical Psychology, 46, 139-149.
Zuckerman, M., Kolin, E. A., Price, L., & Zoob, I. (1964). Development of a sensation
seeking scale. Journal of Consulting Psychology, 28, 477-482.
Zuckerman, M., Kuhlman, D. M., & Camac, C. (1988). What lies beyond E and N?
Factor analyses of scales believed to measure basic dimensions of personality. Journal
of Personality and Social Psychology, 54, 96-107.
Zuckerman, M., Kuhlman, D. M., Joireman, J., Teta, P., & Kraft, M. (1993). A
comparison of three structural models for personality: The big three, the big five, and
the alternative five. Journal of Personality and Social Psychology, 65, 757-768.
Zuckerman, M., Kuhlman, D. M., Thornquist, M., & Kiers, H. (1991). Five (or three)
robust questionnaire scale factors of personality without culture. Personality and
Individual Differences, 12, 929-941.
Chapter 2
Hans Eysenck's personality dimensions: Their
number and nature

C. R Brand

1. TWO DIMENSIONS OF PERSONALITY: N AND E

For a quarter of a century (e.g., Eysenck, 1947, 1972), the name of Hans
Eysenck was associated with commitments to British empiricism, American
behaviorism, and just two dimensions of variation in personality. Neuroticism-
Stability (N) and Extraversion-Introversion (E) were variations that had
respectively fascinated Freud and Jung; but Eysenck would try to furnish a
rigorous, quantitative, scientific account of them. Avoiding commitment to
specific instincts, forces of repression or any notion of a dynamic unconscious
mind, Eysenck would distance his own theory from those of psychoanalysis and
the projective testing movement—though flirtations with astrology and "psi"
would provide reminders that his delight in challenging unmerited authority
was far greater than any boring tendency simply to dismiss engaging or unlikely
ideas. Hans Eysenck's claim would be that N and E were the main well-
established dimensions of individual differences in behavior and that—
mediated by psychological mechanisms—they arose from underlying physio-
logical variations.
Eysenck's commitments would provide a powerful interaction effect that
would singe the beards of philosophers, behaviorists, and personologists alike.
On the one hand, Eysenck was insisting that personality could be studied
scientifically—even in part by the elementary method of just asking people
about their behavior and feelings. He would ride roughshod over classical
objections that a "person" is not properly approached and measured with a
ruler not even pretended to capture conscious experience of "what it is like to
be them"; and over psychoanalytic objections that self-report questionnaires
must have drastic limitations arising from the operation of self-deception and
unconscious motivations. Yet, at the same time, Eysenck was to prove a false
18 Personality

friend to the more radical forms of behaviorism. He did not see individual
differences as well described in terms of hundreds of thousands of behavioral
bits and pieces, each reflecting individual vicissitudes of circumstance and
conditioning. Rather, Eysenck was content from the first to envisage tracing
surface differences—whether in self-reported behavior, tested abilities,
attitudinal preferences, or demonstrated psychopathology—partly to deeper,
underlying dimensions of personality. These dimensions would be indicated by
factor analysis, and they would be explicated scientifically by extrapolating
from experimental work on the laboratory rat.
Thus, from the first, Eysenck's approach as a personality theorist spanned
the usually large divides to be found within psychology—divides that have
often seemed bound to lead to the break-up of psychology altogether. Eysenck
would insist on the possibility of a scientific approach to big questions about
human personality. Unlike today's "social" psychologists, he would not deny
the usefulness of permissible science, and, unlike today's "cognitive" psycho-
logists, he would not focus science only on what are frankly the least interesting
questions about people. Eysenck's faith was essentially that the empiricism and
the correlational methods that had yielded IQ testing and the Spearman
intelligence factor g should be capable, if conjoined with animal experimenta-
tion of clearing more of the jungle of conventional personology. In particular,
he hoped that personality dimensions would soon turn out to reflect
experimentally identifiable individual differences in "conditionability" of one
kind or another. Thus he would "reduce" personality differences to more
psychologically basic ability differences. Additionally, he saw that the same
psychogenetic methods that were allowing Sir Cyril Burt and James Shields to
talk of a substantial heritabililty for IQ differences could be applied to other
psychological dimensions (as to some extent they had been already in German
studies of the role of nature and nurture in crime and schizophrenia).
Although himself a key figure in the development of behavior therapy in
Britain, Eysenck was thus on a collision course not only with psychoanalysis but
also with behaviorism and much of the rest of personology.
Eysenck focused until around 1972 on just N and E and this was a big help to
his mission. A campaign for unification in science is a pleasant sight to all those
who wish to go further than celebrate artistically the complex richness and
nonquantifiable individuality of human beings. To psychology students, N and
E had the attractions of the two-dimensional, North-South/East-West layout
that could be linked to the classic Galen-Wundt circle (see Figure 2.1).
More seriously, Eysenck invariably made an impressive job of arguing that
these two roughly independent dimensions embraced most of the psychometric
variance that could be found with reliability in factor analyses of data from
people who had normal occupational levels and IQs and who were responding
to items including a reasonable number of Eysenck's own. (Frequent subjects
 Hans Eysenck's personality dimensions: Their number and nature 19

aggressive
excitable
changeable
impulsive

optimistic
active

passive sociable
careful
outgoing
thoughtful
talkative
peaceful
responsive
controlled
reliable
even-tempered
calm

Figure 2.1. The Galen-Wundt-Eysenck circle of personality and temperament.

were London bus drivers and housewives recruited commercially to ascertain

their reaction to trial advertisements.) Empirically, N and E seemed plausible
candidates for being "the two major dimensions of human personality."
Finally, with N and E, acolytes were given something else. Especially
attractive were the links that Eysenck made to the only theory that remained to
state-funded psychology after 1945, "learning theory." Largely setting aside the
classic but thorny question (especially after two World Wars and the Holo-
caust) of what actually were the major human drives, Eysenck hypothesized
that N reflected deep-seated motivational "activation" of virtually any kind;
while E-I differences concerned control mechanisms—with introverts being by
their conditionability more readily socialized into control, especially perhaps of
such potentially antisocial drives as sex and aggression (though Eysenck would
mention these drives more than measure them). Here, then was a story to
which many psychologists could relate. As if picking the best of Freud and
Jung, Eysenck had brought in a prosaic, measurable form of "introversion" to
do the job of braking the passions to which Freud had assigned the mysterious
20 Personality

and savage superego. By 1963, Eysenck felt there was enough evidence from
rats to propose an authoritative and catchy characterization of human crim-
inality: recidivist criminals could be seen as relatively unconditional
creatures of their surging passions.
This bold commitment to a theory of crime proved crucial to the develop-
ment of Eysenckian theory, for it soon turned out that at least the more serious
adult criminals could not be shown to be either extraverted or unconditionable
(Burgess, 1972; Passingham, 1967—the latter perhaps unsurprisingly, since
they could after all use knives (if not invariably forks)). Further, under the
research spotlight, it turned out that introversion itself had no really strong
links to measurable conditionability (White Stephenson, Child, Sandra, &
Gibbs, 1968; Kelly and Martin, 1969). Thus the quiet sobriety—not to say
grayness—of the introvert had to be attributed to higher arousal, higher
arousability, inability to reduce arousal, or other arousal-invoking notions that
lent themselves to no very certain measurement. (A high peak of achievement
had been the finding of a .70 correlation between introversion and uncon-
ditioned salivation to lemon juice (Corcoran, 1964), but this finding proved
neither especially intelligible nor readily replicable (Wardell, 1974).) Even
Eysenck's moves to replace "conditionability" with versions of "arousal" were
insufficient to maintain any clear impression of the superiority of introverts.
Far from introverts being conspicuously conditionable or even attentive in the
laboratory, it turned out that it was extraverts who were actually somewhat
better at most laboratory tasks (M. W. Eysenck, 1977). Moreover, CattelTs
researches (M. D. Cattell & R. B. Cattell, 1969; R. B. Cattell & Kline, 1977)
continued to show "Exvia" (Eysenck's Extraversion) and "Control" (today's
"conscientiousness") as independent rather than opposed characteristics.

2. ADDING A THIRD PERSONALITY DIMENSION: P

Could some other human variation realize the dream that had ended with the
failure of extraversion to explain crime—and with the more general failure of
"conditioning" approaches to human behavior and experience by the mid-
1970s? Looked at coolly, Eysenck had plenty of resources upon which to fall
back. By 1954, he had become noted for his recognition of the two-dimensional
attitudinal space remarked by the Thurstones (Thurstone, 1934) and Ferguson
(1944), and for his willingness to identify "extremists" (Fascists and Com-
munists) as equally low scorers on the Tender-mindedness (T) dimension that
cut orthogonally across the Radicalism vs. Conservatism dimension of social
and political attitudes (C). (This precocious appreciation of the psychology of
Communism was probably what had Eysenck blackballed by the British
Establishment and never offered a knighthood.) Furthermore, through the
1960s, Eysenck had gradually allowed that, except in extreme circumstances,
 Hans Eysenck's personality dimensions: Their number and nature 21

where a testee was trying to "fake good" to impress an employer, the "Lie" (L)
scale that he had supplied with the Eysenck Personality Inventory was itself
picking up genuine personality variance in socialization and conscientiousness.
Finally, in his defense of Arthur Jensen after 1969, Eysenck had become more
conspicuously committed to g as a major cause of human variations in
educability, relationships, and attainments. (Having at first broken with Burt
and moved toward clinical psychology, by 1969 Eysenck was forcibly reminded
by the Jensen affair of the importance of the g factor and played an important
part in making Jensen's work known in Britain (Eysenck, 1971,1973; Eysenck
vs. Kamin, 1981).)
Funnily enough, six psychometric dimensions were what R. B. Cattell and
his followers reported they usually found reliably in reasonably catholic item
arrays (Cattell, 1973; Gorsuch & Cattell, 1967). From his internationally
collected data and state-of-the-art factor analyses, Cattell talked of six
independent "second-order" dimensions. So Eysenck had the option of
joining forces in what by then was a de facto empirical agreement between the
two great psychometrician-psychologists (see Table 2.1).
Admittedly, Cattell persistently preferred to see his full 57 "oblique,"
intercorrelated factors as the truly causal agencies behind his "Big 6"
dimensions; Yet Cattell had made no progress towards proving this thesis.
What really alienated Eysenck was probably Cattell's willingness to embrace
Freudian concepts to characterize some of the Cattellian "primaries" (notably
Q4 [id], C [ego] and G [superego]). Faced with the option of sharing with Cattell
six well-established psychometric dimensions which Eysenck's own work and
reading had clearly acknowledged piecemeal, Eysenck instead took up a
position from which his powerful force-field would have unique influence on
the rest of psychometric psychology. He was certainly committed to offering an
understanding of criminality, and of the attitudinal tough-mindedness that he
and other psychologists could readily agree to deplore in Fascists, Com-
munists, and any other political extremists. Yet, though his interest in other
personality dimensions was sharpened, he had no wish to slide down the
slippery slope towards Cattell's multifactorialism and Freudianism.
Instead of ditching behaviorism and linking up with Cattell to back their six
big dimensions, Eysenck embarked on a remarkable personal quest. This was
for another dimension that might at once help explain psychopathy, psychosis,
and political realism—three cognitive-behavioral clusters which had
unpromisingly slight connections with each other. In the 1970s, Eysenck
[now very much "the Eysencks" (H. J. Eysenck & S. B. J. Eysenck, 1977) and
chapter 6] resurrected a psychometric dimension of perceptual and motor
bizarreness and awkwardness that had once been observed to distinguish
psychotic patients (Eysenck, 1952; Eysenck, Granger, & Brengelman, 1957).
Self-report items were written to tap Psychoticism (P)—notably asking about
"paranoid" and bizarre ideas—and Eysenck was soon able to announce that
22 Personality

the dimension successfully distinguished both criminals and psychotics, and

provided the philosopher's stone as to what schizophrenia and manic-
depressive psychoses had in common. With Claridge (e.g., Claridge,
Robinson, & Birchall, 1985; chapter 17) and Jeffrey Gray (e.g., Gray, Picker-
ing, & Gray, 1994) in courteous attendance, Eysenck was able to progress by
the 1990s towards linking the "psychology" of P to a failure of "latent
inhibition" (De La Casa, Ruiz, & Lubow, 1993) of irrelevant stimuli and
memories, and eventually Eysenck (1995) felt able to put P forward as a key
ingredient in any understanding of the psychology of genius—an outstanding
task that had long defeated twentieth-century psychologists (especially those
who could not even bring themselves to mention IQ). Undoubtedly, this was a
major coup (see chapters 6, 19, and 20). Like the Scottish psychiatrist, Laing
(e.g., 1965), Eysenck had linked the major forms of "deviance" together and
provided a psychology and a rationale of how the less agreeable forms of
deviance can persist in human populations. Simply, the same factors that yield
genius and valor are those that, in other (admittedly unspecified) interactions,
yield psychopathy and madness. Coupling this mighty personality dimension
with a new belief that brief psychotherapy could change personality (and spare
people from cancer), Hans Eysenck was able to take on all comers even in the
depths of his retirement. To his fellow academics, he tirelessly defended his
"Big 3" against the by-then-fashionable "Big 5"; and large crowds of
psychology students would happily turn out to hear him when he talked of
the life-saving psychotherapy that had been developed and tested in
Yugoslavia (e.g., Eysenck, 1993).
Not everything in the Eysenckian garden was so rosy, however. Persistent
problems with P were as follows.
1. P turned out repeatedly to have a markedly skewed distribution, sug-
gesting that it must be—like crime and schizophrenia themselves—the
product of multiplicative interaction effects and not a basic personality
dimension in its own right (e.g., Block, 1977).

Table 2.1. Approximate correspondence between the main independent dimensions identified by
Eysenck and Cattell around 1970

Eysenck Cattell
1. Neuroticism Anxiety (id [tension], guilt, fear, —ego)
2. Extraversion (sociability, impulsivity) Exvia (surgency, affectothymia, assertiveness)
3. "Lie'Yhypocrisy/socialization Control (superego, self-sentiment)
4. Radicalism Independence (assertiveness, bohemianism,
radicalism, self-sufficiency, ego)
5. Tender-mindedness/idealism Pathemia (tender-mindedness, affectothymia)
6. General intelligence General intelligence
 Hans Eysenck's personality dimensions: Their number and nature 23

2. P's correlations with the variables in its "package" were seldom

particularly strong: psychopaths and psychotics were only about one
standard deviation high, whereas these people are in a remote 1% of the
population by ordinary clinical and human standards. Moreover, the many
people who have only psychotic depressions (without the manic swing)
have never been particularly high-scoring on P at all. Nor did the other
components show any systematic sign of correlating among themselves:
schizophrenics have an entirely normal crime rate, and, looking across the
board, the generality of the evidence has never really favored Eysenck's
idea of schizophrenia and manic-depression running in the same families
(Van Kampen, 1996). The attempt to link P to tough-mindedness was not
much more successful: in the largest study (of 808 children), P correlated
only around —.30 with religiosity and around +.20 with ethnocentrism and
punitiveness (Powell & Stewart, 1978).
More promisingly, P has correlated at -.75 with degree of dopamine
binding in the basal ganglia (Gray et al., 1994)—but that study only
involved an TV of 9.
3. Attempts to provide a psychology of P took a meandering path before
recently halting at "latent inhibition." For a while P was linked to
aggression; then "tough-mindedness" was tried (e.g., Wilson, 1978); then
lack of conscientiousness was ventured (the term "low superego" was even
used by the Eysencks occasionally); and modern workers invariably find P
linked to "disagreeableness" (Brand, 1994) [i.e., to what Cattell had less
evaluatively measured as Independence (or "Promethean Will")]. In fact,
one of the higher correlations ever reported for P was of —.46 with Verbal
IQ (in 103 patients at Rampton Special Hospital—Davis, 1974).
4. Needless to say, the psychology of genius furnishes little but anecdotal and
sketchy evidence of high psychoticism. Moreover, Eysenck has had to
plunder evidence that could equally be taken to support the idea that genius
was linked rather to neurosis rather than psychosis: evidence of alcoholism,
sexual fixations, and suicide in writers, for example, is as easily taken as
evidence of high neuroticism as of high psychoticism. A particular problem
is that many geniuses are tender-minded and very hard working [as Eysenck
(1996a) has himself reproached me for ignoring]; yet P is, if anything,
plainly associated with low tender-mindedness and low conscientiousness.
Eysenck's attempt to introduce "P versus niceness" to personality study
resembles Adler's attempt to give a large role to people's efforts to overcome
their inferiority feelings, sometimes resulting in undue egoism and hostility to
others. Eventually, Adlerians had to agree there was more than one dimension
of "egoism versus altruism" (Ansbacher, 1985) and Eysenck will probably find
that he has to do the same. The jury is still out, and P must have a sporting
chance while the "Big 5" remain without any psychological or physiological
24 Personality

account whatsoever (e.g., McCrae & Costa, 1996). Yet delays in developing a
psychology of the Big 5 may themselves reflect incorrect appreciations of what
are the main dimensions of human personality for which accounts must be
found.

3. THE ROLE OF INTELLIGENCE AND MATURATION

The consequences of Cattell and Eysenck failing either to beat each other
decisively or to compromise around 1972 have been enormous. Additionally,
the problem for others of choosing between these giants has been made worse
by the internal divisions within each of the two camps. On the one hand,
Cattellians were torn between preferring Cattell's well-established Big 6
dimensions or his 16 less reliable, intercorrelated yet theoretically causal
"primaries." On the other hand, Eysenckians hardly knew whether to back the
Big 6 that had always been present in Eysenck's work (or seven if P had to be
added instead of being an interaction effect between some of the others) or to
join up with Eysenck's insistence on his "Gigantic 3" (viz. P, E, and N) (e.g.,
Eysenck, 1991). Left in the lurch by the failure of any clear solution to emerge
from the two giants of differential psychology, pygmy psychometrician
psychologists proceeded to lose their grasp of both Eysenck and Cattell.
Seeking factorial truth, like Cattell, they looked for dimensions, but they
usually looked in highly convenient databases that seldom involved lower-IQ
subjects, and—making doubly sure of error—they even abandoned Cattell's
practice of always including IQ-type items alongside self-reports of habits and
interests. Thus they tended to miss the g factor which itself seems to make for
more independent character traits at its higher levels (Brand, Egan, & Deary,
1994). Moreover, the fact that Eysenck and Cattell did not agree with each
other made outright newcomers to the scene susceptible to the idea that the
truth must lie "somewhere in between" the giants' more familiar positions. By
outlawing g from the realm of personality, they achieved just this result: a Big 5
that fell between Cattell's six-or-more and Eysenck's Gigantic 3 (Brand,
1994a). On the other hand, if g is included—as it tends to be thanks to
respondents themselves in studies of "person perception"—a very broad,
evaluative, "Disagreeableness" dimension (cf. Warr & Knapper, 1968)
smothers the distinction between independence and tough-mindedness: these
two separate, independent dimensions emerge more readily when raters know
ratees especially well, and notably when respondents are describing themselves
(Brand, 1994b). The result is that Big 5 theory (some talk of the "Five Factor
Model," but in reality there is no "model" provided of anything in any
scientific sense) is also a house perpetually divided against itself: studying the
high-IQ yields one solution, and studying the person-perceptions of more
normal people yields another!
 Hans Eysenck's personality dimensions: Their number and nature 25

Still more serious than losing (by one method or another) a whole dimension
of psychological variation from the six that Cattell and Eysenck had once had
de facto in common, Big 5 theorists have also—by their failure to read and to
ask why Eysenck and Cattell obtained the different results that they did—
missed vital clues to the structure of human personality and the processes by
which it develops. More especially, they have failed to link personality dimen-
sions to typical levels of mood and motivation—the kind of job that Eysenck
had long had in mind when treating personality largely as if it could be reduced
to temperament. This is a singular failure in view of the obvious connections
between terms for mood, emotion, and temperament (depressed/serious,
fearless/confident, hostile/cynical, etc.) The failure of today's psychometrician-
psychologists to look at what happened historically means that their continuing
work today provides a frightening demonstration of that tendency to "factor
analysis for the sake of factor analysis" about which Eysenck always
complained.

4. DOES THE NUMBER OF PERSONALITY DIMENSIONS DEPEND ON THE

SAMPLE STUDIED?

Since the internally divided Big Fivers cannot be right, it is time to entertain
seriously the possibility that Eysenck and Cattell might themselves both have
been right; and especially that they were right in what they both asserted and
wrong in what they separately denied. In what follows, the possibility will be
examined that the number of major dimensions does truly vary somewhat
depending on the sample—though six (if Eysenck's P is presumed an inter-
action effect) will usually be seen in better-educated and more mature subjects
of above-average intelligence. Subsequently, the time-honored insistence of
Eysenck (unlike Cattell) on providing psychology as well as psychometry will
be considered.
Let us take first g and n—the two dimensions that have always been
acknowledged not just by Eysenck but by his London School forebears (e.g.,
Schonell, 1948, pp. 21-22) and all modern psychometrician-psychologists who
are not so politically correct as simply to avoid reference to IQ in their work.
These two dimensions reflect classic conceptions of Reason and the Passions
and their wide-ranging contest and collaboration. They can both be readily
measured—with a little allowance for test sophistication and lying; and they
both have plausible psychological underpinning in, respectively, information
intake and information retention (Brand, 1996, chapter 2; Brand, 1997). As
Burt and the London School always maintained, these are surely the most
important ways in which people differ, and they have intrinsic links to
considerations of value. There is no merit in not being able to take in new
information—even if such inability helps preserve traditional moral belief.
26 Personality

Information-retention is likewise presumably meritorious in principle—though

memories can be useless and burdensome, as well as useful, and the inability to
excise or rewrite memories arguably results in neurosis (as Freud first
envisaged).
Interestingly, differences in both g and n probably have connections to broad
and wide-ranging mood features (e.g., Brand, 1994b). High-g people tend to be
experienced as more alert and receptive, and high-n people are more variable
in moods and behavior. But these broad emotional features probably obtrude
rather little into immediate conscious experience. People's more immediate
experiential concerns are probably with the short-term, limited, and resoluble
conflicts involving more closely targeted emotions and their restraint, as
follows.
Considering first the tensions that have led Eysenck himself to invoke the
name superego, how could Eysenck for many years have thought extraverts
crime-prone when even his own evidence from the E and L scales showed little
sign of the negative correlation that such theorizing should have led him to
expect? True, by 1972, he had come to note the connection between E and
sexual vigor and frequency, and to admit the special independent status for L
that proved a harbinger of modern Big 5 measures of "conscientiousness" (vs.
impulsiveness/casualness). Yet, once upon a time, S. B. G. Eysenck and H. J.
Eysenck (1963) had insisted to an astonished American psychometric world
that the sociable and impulsive "facets of extraversion" were correlated at
+.47, compatible with the "unity" of Extraversion. One has to ask: Which is
right? And one has to answer: Both views are right, but the cruder, broader
extraversion-versus-conscientiousness dimension emerges more easily in full-
population samples like Eysenck's rather than in the higher-IQ groups that
furnish most of the data for the typical modern Big 5 theorist. For years it was
thought that the Baltimore sample that yielded the Big 5 had a reasonably
normal collection of subjects, but eventually it transpired that no less than 25%
of McCrae and Costa's subjects actually had Ph.D.s! Looking at all this
objectively, one has to say that extraversion and conscientiousness are more
identifiably separate at higher g levels—and (in Edinburgh pilot researches) at
lower levels of neuroticism, whether as cause or effect. If there is a tension
between energy and conscience, or between sexuality and superego in
particular, it goes on more keenly (or more visibly) in lower-g, higher-n people.
Now, if a crude contrast of energy versus inhibition perhaps differentiates in
the course of intelligent maturation into two distinct dimensions having no
marked or simple influence on each other, differential psychologists will ask
for parallels and analogies that may help us to understand such processes.
Plainly some kind of Freudian story is a possibility: that the initial stark
opposition between eros and superego becomes—during normal, healthy
development—less of a conflict than two distinct sources of strength. But
most differential psychologists will be reminded of another dimension that
 Hans Eysenck's personality dimensions: Their number and nature 27

differentiates over time, and especially at higher-intelligence levels. This is the

dimension of g itself: g increasingly yields a distinction between "fluid" and
"crystallized" intelligence as aging proceeds, and especially when g{ itself goes
into decline after age 55. Following this analogy, people could vary
independently in energy and conscientiousness so long as they are of
relatively good intelligence and maturity; yet at lower IQ levels a more stark
contrast might be observed between rambunctious extraverts and controlled
introverts in the broad sense of extraversion envisaged by Eysenck back in
1963. With maturation (and higher g), the broad emotional contrast between
exuberance and restraint crystallizes to allow more distinct dimensions of
elation versus depression (extraversion-introversion) and obsession/caution
versus hysteria/excitability (conscientiousness-casualness).
A similar dependence of dimensionality on maturation may affect the other
great emotional tension between exhibiting a strong and indeed aggressive
identity on the one hand and being more accepting, tender-minded, and open-
minded on the other. Here, the Freudian (or Kleinian) story would be that an
early death wish allowing destruction of useless parts of the self—whether of
cells or of psychological systems—may be harnessed by the self and turned into
useful aggression; or it may itself be destroyed by the self, leaving a pacific,
agreeable, and less egoistic yet less directed individual (programmed cell death
is known as apoptosis). It was realized in the nineteenth century (and known to
Freud) that the body has self-destruction mechanisms that come into action
unless cells receive particular patterns of messages from other cells, (C.
Badcock, personal communication). However, a non-Freudian differential
psychologist may well settle for a story of differentiation whereby the crude
disagreeable-versus-agreeable distinction becomes modified according to how
much crystallization takes place into two independent variations of inde-
pendence versus dependence and tough-versus tender-mindedness. Plainly the
original primitive tension here is whether to allow or reject hostility, and that is
the kind of contrast made by Eysenck's P scale—though P itself probably
involves a multiplicative interaction with low conscientiousness also, thus
helping to yield its skewed distribution. What characterizes the high-will and
high-aggression person is an awesome realism about the human condition that
is alert to the need for competition and inclined to disregard the possibilities of
cooperation and idealistic constructivism. Yet, as both Freud and Adler
recognized, that crude contrast will diminish with maturation. In terms of
emotion, the two separate dimensions that emerge are probably those of anger
versus fear (as the fluid form of independence vs. dependence) and disgust
versus acceptance (as the fluid form of tough- (or nonopen-) mindedness).
Pilot research in Edinburgh has in fact suggested these dimensions to be more
separate at higher levels of neuroticism (where ego development has perhaps
been less decisively in the agreeable or disagreeable directions).
28 Personality

5. THE "BIG 6"

The above arguments and descriptive considerations yield some six major
claims, as follows.
1. The number of major personality dimensions found will vary between
some four and six, depending on the intelligence of subjects and on
whether the self-report questions succeed in tapping intelligent crystal-
lizations of temperament.
2. Eysenck's once broad conception of extraversion as involving liveliness,
sociability, and impulsivity typically breaks up in better-educated subjects
into two independent dimensions of extraversion and conscientiousness.
Likewise, Eysenck's P (and the Big Fivers' "Disagreeability") can be
broken into components of Cattellian Independence, Pathemia, and lack
of Superego. (For perhaps the first claim to detect such "now-you-see-it-
now-you-don't" differentiation, see H. J. Eysenck & M. W. Eysenck
(1985), Figure 12, p. 119: "The hierarchical structure of the affective
system.")
3. Differentiation of dimensions may be driven forward by processes of
crystallization—just as for the intelligence dimension itself. Each dimen-
sion may be considered to allow somewhat separate crystallized forms—in
the realm of attitudes, say, as distinct from the more fluid and untutored
realm of temperament. Alternatively, it might be that the cruder, broader
contrasts that are sometimes seen reflect the fact that it is easier to read
the final outcome of Freudian battles than to say much about the precise
strengths of the competing players (as in "eros vs. superego" and
"thanatos x ego").
4. The full six dimensions might be summarized as follows—using a catholic
array of descriptors but highlighting those that I have myself inclined to
work with over the past decade (see Brand, 1984).
• Intelligence, general intelligence vs. concretistic thinking
• Emotionality, neuroticism, sensibility, id vs. placidity, stability, sense
• Extraversion, energy, surgency, eros vs. introversion, gravity
• Conscientiousness, control, superego vs. impulsivity, casualness,
liberality
• Disagreeableness, will, independence, ego vs. subduedness, passivity,
affability
• Openness, Culture, affection, idealism vs. tough-mindedness, cynicism,
thanatos
5. A more specific linking of the last four of the above dimensions to specific,
keenly experienced dimensions of mood and affect might be achieved as
follows.
e+ vs. e— = elation vs. depression, sadness, seriousness
 Hans Eysenck's personality dimensions: Their number and nature 29

A radicalism will ego

L C
T O
E N
R S
A E
T R
1 conscience/ 'lie' V
0 energy/eras A
control/ 'superego
T
1 1
extraversion thanatos defensiveness S
s
M M

tender-mindedness
affection

Figure 2.2. Four dimensions of personality (other than g and n) shown with regard to how they may
sometimes appear as fused and how they may relate to social and political attitudes.

c+ vs. c— = vigilance vs. fatigue, boredom, casualness

w+ vs. w— = confidence vs. fear, submission, deference
a+ vs. a— = friendliness vs. hostility, suspicion, cynicism
Such linkage makes use of the major dimensions of mood and emotional
variation that are identified in modern researches (Brand, 1994b).
6. The way in which the last four personality dimensions fuse at lower IQ and
become more separate at higher IQ is illustrated in Figure 2.2, as is their
likely relation to dimensions of social and political attitudes.

6. THE PSYCHOLOGICAL MEANING OF THE "BIG 6"

Even if Eysenck and Cattell were thus both right as and when they recognized
six dimensions—some of which can perhaps allow differentialists to see how
Freud came by some of his own ideas—the above propositions still do little to
address questions of psychological explanation on which Eysenck has always
insisted.
30 Personality

How should the Big 6 be understood in psychological terms? Sadly, despite

many years of research—especially into extraversion—the picture is still very
unclear (e.g., Gale & M. W. Eysenck, 1992; Matthews, 1993). Here are some
possibilities that still look viable yet falsifiable.
1. g is the efficiency with which even quite basic (perceptual) information is
processed (e.g., Brand, 1996, chapter 2).
2. n is variability on any or all of the other dimensions of personality. The
emotional extremes of higher n may allow better long-term subcortical
storage of personal memories—of events and episodes (e.g., Brand, 1996,
pp. 48-49; Brand, 1997).
3. e involves the ability to reduce levels of conscious arousal, whereas
introverts become easily overloaded by noise, demands for speed, and the
presence of others. Thus, extraverts cope well with immediate laboratory
tasks requiring vigilance, attention, persistence, and short-term memory,
while introverts do better when testing is extended over several days and
positively requires long-term memory storage (Brand, 1994b; Matthews,
1993).
4. c involves the ability to sustain arousal—allowing perseverance and
preventing boredom, casualness, impulsiveness, excitability, and negli-
gence (see Matthews, 1993). Maintenance of lateral inhibition is probably
involved here—as in Eysenck's low-P.
5. w involves the ability to attend narrowly, in a focused manner, avoiding the
influence of the perceptual field to some degree. The related dimension of
"field-independence" was explored particularly in the work of the USAF
psychologist Herman Witkin (see Brand, 1996, pp. 45-46).
6. a, by contrast, seems linked to the ability to process a rather wide range of
abstract information with the help of imagination and high-level cultural
symbols and ideas—or of what has lately been called an associational
network of symbol manipulation (see Brand, 1996, pp. 43-45). Higher-a
people are less practical and realistic, and more trusting and idealistic
because of being at ease in the worlds of symbol systems. (Higher a is more
common in women, just as higher w is more common in men.)
A further suggestion would be that each of the last four dimensions
represents a degree of "break-out" from two important "trade-off functions"
that are found widely in psychology. One is the speed versus accuracy trade-off,
which, if uncorrected, makes for opposition between e and c (rather than for
the independence that is more commonly observed in real-life tasks carried out
by intelligent adults). The second is the trade-off between narrow versus broad
attention: without correction by intelligence and learning, this trade-off would
yield an opposition between w-like and a-like features, rather than the
independence that is more commonly found in personality data (see Brand,
Egan, & Deary, 1993).
 Hans Eysenck's personality dimensions: Their number and nature 31

In overview of the above Big 6 and their psychology, the suggestion is thus
that g is a dimension of information intake, that n is a dimension of infor-
mation storage, and that the other four dimensions are all concerned with the
modulation of levels of conscious arousal—with e and c making adjustments
affecting behavioral output, and w and a modifying attention and intake (see
Brand et al, 1994; Brand, 1994b, 1996). All these speculations are doubtless
premature, but the "converging consensus" of psychometricians will surely
soon unleash kindred efforts to complete the task of understanding
psychologically dimensions which Eysenck first identified in the 1950s and
began to study experimentally in the 1960s.

7. EYSENCKIAN PSYCHOLOGY IN PERSPECTIVE

Whether the pursuit of such hypotheses will disclose the "bases" of personality
is as fascinating a question as when Eysenck first linked Neuroticism to general
instinctual midbrain activity and Introversion to the inhibitory workings of the
cortex around 1960. Unlike Freudian ideas of dynamic and invisible conflicts
(as the ego and the superego wrestle to harness major drives), the above
hypotheses do border on being testable and as such are in line with Eysenckian
philosophical demands. Yet whether such searches for bases will yield much
more than previous mechanistic hypotheses about personality must remain a
moot point.
There may be real problems for mechanistic approaches if they are expected
to walk off with major slices of big personality dimensions. For the past 20
years, g itself has shown strong and replicable correlations with basic processes
such as "inspection time" (especially across normal levels of g—Brand, 1996).
Yet few differential psychologists—even Eysenckians—have been tempted to
declare that we now know "what intelligence really is." Again, even if Neuro-
ticism turns out to be strongly linked to "personal event memory" (Brand,
1997), who precisely will think such identification is a giant leap forward for the
psychology of personality? There is perhaps a problem as to how an Eysenck-
ian reduction can ever be completed—even to the satisfaction of Eysenckians!
The agenda of Eysenckianism looks still less satisfying if it is asked why such
theorizing has had so little to offer about the real mysteries of mankind. What
is the evolutionary function of different people having the markedly different
sexual orientations and styles that they do? Under what special circumstances
and in conjunction with what other variations are particular sexual
idiosyncrasies (frankly, "perversions") functional? Why are some people
masochistic? Why do some people self-mutilate and attempt to commit
suicide? Try as one might, it is hard to envisage what might be the connection
between the variables of Eysenckian psychology and any immediate answer to
such questions.
32 Personality

Of course, Eysenck can properly reply that he has deliberately not tried to
answer all the questions on the examination paper. Instead, he has shown the
possibilities of science helping with some questions—and never more than with
the questions of how best to "carve nature at the joints" and identify the
origins of the main human personality differences (seldom environmental as
these have crucially proved to be). Practically, he can point to the enormous
encouragement his approach has always given to those who have sought to
challenge the conception of people as pawns of environmental forces, and to
those who have sought tangible psychological improvements, whether by drugs
or behavior therapy. Yet the truth is that he could have had more luck with his
psychology. Like Freud himself, Eysenck may have succeeded in getting across
a broad conception of human variation, the precise bases of which are actually
very hard to pin down. It has to be said that Eysenck and Eysenckians have still
left the field open to Freudian psychological concepts—even if, like Freud
himself by 1930, they have doubted the usefulness of psychotherapy.
Nothing could be more certain than that Hans Eysenck does not regard
himself as a Freudian. He first shot to fame in 1951 with his classic
demonstration of the inefficacy of psychoanalysis, and in recent years he has if
anything redoubled his attack (e.g., Eysenck, 1991, 1995, 1996b). Nor could
Eysenck ever be a convinced or convincing Freudian: his demand for empirical
evidence is such as to exclude scientific talk of unconscious conflicts and
repression (see Brand, 1993). Rather than envisage people having sexuality in
infancy or a death wish at any age, Eysenck (e.g., 1983) would rather talk of the
influence of the stars on personality—so much better was the evidence for the
Gauquelins' astrological theory than for Freud's. Essentially, Hans Eysenck is
not a Freudian.
Nevertheless, Eysenck's own times have not witnessed any triumph of the
relatively mechanistic psychology that he himself espoused—except insofar as
many cognitive scientists now seem to spend more time studying how to make
robots than how to understand people. Perhaps one day an Eysenckian agenda
of the type outlined above will be completed—and perhaps even a new P with a
normal distribution will be revealed in all its glory. However, until that time,
Eysenck should take quiet satisfaction from the following Une of thought. It
might still turn out that the scientific bases of personality are (as Cattell
thought) scattered among the "primaries"; and that (as William McDougall
thought) the major dimensions of psychological variation require accounts that
link them to individual dynamics, purpose, and biological function. Yet,
alongside Eysenck's many achievements in relating personality to psycho-
pathology and politics and in countering egalitarian and social-environ-
mentalist fables of human development, Eysenck's traditional six dimensions
place him head and shoulders above the inconsistencies and unimaginativeness
of the present Big 5. They leave him a solid record of achievement (together
 Hans Eysenck's personality dimensions: Their number and nature 33

with Cattell), a special reputation for insisting on psychology (and not just
psychometry), and the option of being considered, in history's book—despite
himself at present—a fair-minded, empirical, accidental Freudian.

REFERENCES

Ansbacher, H. L. (1985). The significance of Alfred Adler for the concept of narcissism.
American Journal of Psychiatry, 142, 2, 203-206.
Block, J. (1977). The P scale and psychosis: continued concerns. Journal of Abnormal
Psychology, 86, 4, 431-434.
Brand, C. R. (1984). Personality dimensions: an overview of modern trait psychology. In
J. Nicholson, & H. Beloff (Eds.), Psychology Survey 5 (pp. 175-209). Leicester: British
Psychological Society.
Brand, C. R. (1993). Special review of H. J. Eysenck, Decline and fall of the Freudian
empire, Washington, DC: Scott-Townsend, 1990. Behaviour Research & Therapy, 31,
1, 129-131.
Brand, C. R. (1994a). Open to experience—closed to intelligence: why the "Big Five"
are really the Comprehensive Six. European Journal of Personality, 8, 299-310.
Brand, C. R. (1994b). How many dimensions of personality?—The "Big 5", the
"Gigantic 3" or the "Comprehensive 6"? Psychological Belgica, 34, 4, 257-273.
Brand, C. R. (1996). The g factor: General intelligence and its implications, Chichester,
U.K.: Wiley.
Brand, C. R. (1997). What is it to be high-Neuroticism?—Drive, vigilance or personal
memory? In J. Mohan & H. J. Eysenck (Eds.), Psychology and cross-cultural
perspectives. London: Taylor & Francis.
Brand, C. R., Egan, V., & Deary, I. J. (1993). Personality and general intelligence. In
G.van Heck, P. Bonaiuto, I. J. Deary, & W. Nowack (Eds.), Personality psychology in
Europe: Vol. 4, (pp. 203-228). Tilburg, NL: Tilburg University Press.
Brand, C. R., Egan, V., & Deary, I. J. (1994). Intelligence, personality and society:
"constructivist" vs. "essentialist" possibilities. In D. K. Detterman (Ed.), Current
topics in human intelligence: Vol. 4 (pp. 29-42), Norwood, NJ: Ablex.
Burgess, P. K. (1972). Eysenck's theory of criminality: A test of some objections to
disconfirmatory evidence. British Journal of Social and Clinical Psychology, 11,3,248-
256.
Cattell, R. B. (1973). Personality and Mood by Questionnaire. New York: Jossey-Bass.
Cattell, R. B., & Cattell, M. D. (1969). Handbook for the junior-senior high school
personality questionnaire. Champaign, IL: Institute for Personality and Ability
Testing.
Cattell, R. B., & Kline, P. (1977). The scientific analysis of personality and motivation.
New York: Academic Press.
Claridge, G. S., Robinson, D. L., & Birchall, P. M. A. (1985). Psychophysiological
evidence of "psychoticism" in schizophrenics' relatives. Personality & Individual
Differences, 6, 1-10.
Corcoran, D. W. J. (1964). The relation between introversion and salivation. American
Journal of Psychology, 77, 298-300.
34 Personality

Davis, H. (1974). What does the P scale measure? British Journal of Psychiatry, 125,161-
167.
De La Casa, L. G., Ruiz, G., & Lubow, R. E. (1993). Latent inhibition and psychoticism.
British Journal of Psychiatry, 84, 119-132.
Eysenck, H. J. (1947). Dimensions of personality. London: Kegan Paul.
Eysenck, H. J. (1952). The scientific study of personality. London: Routledge & Kegan
Paul.
Eysenck, H. J. (1971). Race, intelligence and education. London: Temple Smith.
Eysenck, H. J. (1972). Psychology is about people. London: Allen Lane (Penguin Press).
Eysenck, H. J. (1973). The inequality of man. London: Temple Smith.
Eysenck, H. J. (1983). Special review (of Francpise Gauquelin, The Psychology of
Planets. San Diego, CA: Astro Computer Services). Personality & Individual
Differences 4, 6, 712-718.
Eysenck, H. J. (1991). Dimensions of personality: 16, 5 or 3?—Criteria for a taxonomic
paradigm. Personality & Individual Differences, 12, 8, 773-790.
Eysenck, H. J. (1993). Prediction of cancer and coronary heart disease mortality by
means of a personality inventory: results of a 15-year follow-up study. Psychological
Reports 72, 499-516.
Eysenck, H. J. (1995). Genius: The natural history of creativity. Cambridge, U.K.:
Cambridge University Press.
Eysenck, H. J. (1996a). Special review (of C. R. Brand, The g factor: General intelligence
and its implications. Chichester, U.K.: Wiley). Personality & Individual Differences 21,
5, 829-830.
Eysenck, H. J. (1996b). Review (of E. Erwin, A final accounting—philosophical issues in
Freudian psychology. Cambridge, MA: MIT Press). Behaviour Research & Therapy, 34,
7,599.
Eysenck, H. J., & Eysenck, M. W. (1985) Personality and individual differences: A natural
science perspective. New York: Plenum.
Eysenck, H. J., & Eysenck, S. B. G. (1977). Psychoticism as a dimension of personality.
London: Hodder & Stoughton.
Eysenck, H. J., Granger, G. W., & Brengehnan, J. C. (1957). Perceptual processes of
mental illness. London: Chapman & Hall.
Eysenck, H. J. versus Kamin, L. (1981). Intelligence: The battle for the mind. London:
Pan.
Eysenck, M. W. (1977). Human memory: Theory, research and individual differences.
Oxford: Pergamon Press.
Eysenck, S. B. G., & Eysenck, H. J. (1963). On the dual nature of extraversion. British
Journal of Social and Clinical Psychology, 2, 46-55.
Ferguson, L. W. (1944). Socio-psychological correlates of the primary attitudes scales. I:
Religionism. II: Humanitarianism. Journal of Social Psychology, 19, 81-98.
Gale, A., & Eysenck, M. W. (1992). Psychophysiology: A programme for individual
differences research? Handbook of individual differences: Biological perspectives.
Chichester, U.K.: Wiley.
Gorsuch, R. L., & Cattell, R. B. (1967). Second-stratum personality factors defined in
the questionnaire realm by the 16PF. Multivariate Behavioral Research, 2, 211-224.
 Hans Eysenck's personality dimensions: Their number and nature 35

Gray, N., Pickering, A., & Gray, J. A. (1994). Psychoticism and dopamine D2 binding in
the basal ganglia using single photon emission tomography. Personality & Individual
Differences, 17, 3, 431-434.
Laing, R. D. (1965). The divided self. Harmondsworth: Penguin.
Kelly, D., & Martin, I. (1969). Autonomic reactivity, eyelid conditioning and their
relationship to Neuroticism and Extraversion. Behaviour Research & Therapy, 7, 33-
244.
Matthews, G. (1993). Extraversion, arousal and performance. In A. P. Smith, & D. M.
Jones (Eds.), Factors affecting human performance. London: Academic Press.
McCrae, R. R., & Costa, P. T. Jr. (1996). Toward a new generation of personality
theories: theoretical contexts for the Five Factor Model. In J. S. Wiggins (Ed.), The
five-factor model of personality: Theoretical perspectives (pp. 51-87). New York: The
Guilford Press.
Passingham, R. E. (1967). Crime and personality: A review of Professor Eysenck's theory.
Unpublished master's thesis, University of London.
Powell, G. E., & Stewart, R. A. (1978). The relationship of age, sex and personality to
social attitudes in children aged 8-15 years. British Journal of Social and Clinical
Psychology, 17, 307-317.
Schonell, F. J. (1948). Backwardness in basic subjects (4th ed.). Edinburgh: Oliver &
Boyd.
Thurstone, L. L. (1934). The vectors of the mind. Psychological Review, 41,1, 1-32.
Van Kampen, D. (1996). The theory behind psychoticism. European Journal of
Psychology, 10, 1, 57-60.
Wardell, D. (1974). Stimulus intensity and introversion-extraversion. British Journal of
Social & Clinical Psychology, 13, 4, 425^26.
Warr, P. B., & Knapper, C. (1968). The perception of people and events. London: Wiley.
White, J. H., Stephenson, G. M., Child, Sandra E. A., & Gibbs, Jaynce M. (1968).
Validation studies of the Eysenck Personality Inventory. British Journal of Psychiatry,
114, 63-68.
Wilson, G. D. (1978). Introversion/extroversion. In H. London, & J. E. Exner (Eds.),
Dimensions of personality. New York: Wiley.
Chapter 3
Individual differences in reactions to reinforcing
stimuli are neither black nor white: To what
extent are they Gray?

A. D. Pickering, P. J. Corr, J. H. Powell, V. Kumari,

J. C. Thornton, and J. A. Gray

1. INTRODUCTION

Ideas about the biological basis of personality go back at least to the time of the
Greeks and their speculations concerning the determining roles played by the
various "humors." There are still traces of these speculations in the language
("phlegmatic," "choleric," and so on); but the beginning of modern work in
this field surely lies in the studies of conditioning by Pavlov that led to his
typology of dogs (see Gray, 1964). In such a skipping history, which lands only
on the highest peaks, only one name can follow Pavlov's, that of Hans Eysenck.
Both the similarities and the differences between Pavlov's and Eysenck's work
are instructive. Like Pavlov, Eysenck seeks for the basis of personality in
individual differences in the functioning of various aspects of the central
nervous system, and, like Pavlov, he considers those functions that underlie
processes of learning as being of particular significance in this regard. Unlike
Pavlov, however, who extrapolated from the behavior of dogs to that of human
beings and never seriously studied the latter, Eysenck's starting point lies in a
careful analysis of the structure of individual differences at the human level.
His search for an explanation of personality calls only subsequently upon what
is known from studies of experimental animals about the way the brain controls
behavior. Of course, Eysenck was not alone in studying and analyzing the
structure of personality at the human level. If one confines oneself to that
segment of the field, there are other important figures, above all Raymond
Cattell (who emigrated from England to the United States at much the same
time that Eysenck took the journey from Germany to Britain—is being an
outsider looking in, one wonders, a special spur to being interested in per-
sonality?). What sets Eysenck apart from other analysts of human personality,
however, is precisely that he did seek an explanation, and, moreover, a
 Individual differences in reactions to reinforcing stimuli 37

scientific explanation, amenable to the usual routes of empirical inquiry and

experimental testing. Others, by and large, confined themselves to statistical
analysis and description, the hunt for appropriate (or sometimes merely fancy)
names for personality factors taking the place of any true effort after
explanation.
What sets Eysenck aside from his contemporaries in the study of personality,
then, is first and foremost the sheer fact that he sought for testable theoretical
accounts in a field that notoriously lacked them. It would be nice to record that
a flood of other such accounts followed his lead, since science often flourishes
most when there are competing theories distinguished by their empirical
predictions. However, this has not been so: with a few honorable exceptions
(and none that share Eysenck's own all-embracing scope), alternative views
have largely consisted of relatively minor changes in the naming, location, or
interpretation of factors, not in the underlying structure of explanation. And,
where alternative explanations have been advanced, they have all—including
our own—inherited their basic conceptual scheme from the one first adum-
brated by Pavlov and massively developed by Eysenck: a scheme that
postulates a small number of major dimensions of personality, each of which
reflects individual differences in the functioning of a subsystem of the brain
that is defined simultaneously in terms of its role in the control of behavior and
its place in the economy of the central nervous system (Gray, 1972). Within the
general field of personality description, there is now widespread convergence
towards small numbers of major dimensions: Eysenck himself is still at the low
end of the spectrum, advocating 3, as he has done since the 1950s (e.g.,
Eysenck, 1953,1967; H. J. Eysenck & S. B. G. Eysenck, 1975), but the high end
has retreated dramatically from the 26 of Cattell (1957), or the 14 of Guilford
and Zimmerman (1956), to perhaps 5 (Costa & McCrae, 1985) or 7 (Clon-
inger, Svrakic, & Przybecky, 1993). As to the underlying central nervous system
structures and processes that give rise to these dimensions, even at the
currently high end, as in Zuckerman's (1991) well thought-out framework for
the biological basis of personality, the family resemblance to Eysenck's own
suggestions is plain to see.
Eysenck, in short, dominates the study of the biological basis of personality.
Our own work is no exception to this generalization. It started out from the
suggestion (Gray, 1970,1981) that, although the Eysenckian three-dimensional
space for personality description is correct, some of its dimensions may need a
degree of rotation to lie most snugly along the lines of causal influence; that
those lines of causal influence are, as in Eysenck, individual differences in the
activity of separable subsystems of the brain; but that, rather than being
systems mediating different kinds of arousal, they are systems mediating
different kinds of reactions to reinforcing stimuli; and, accordingly, that they
have a different neurology than the one Eysenck proposed. We shall refer to
38 Personality

our approach as reinforcement sensitivity theory (RST). We go on to consider

how this theory has fared, especially in the light of data recently gathered in
our own laboratories.
Over the decades since RST was first proposed, a steady growth of empirical
studies relating to this theory has occurred. One might attempt some kind of
overall appraisal of the varying degrees of support conferred to RST by these
studies, as we have done in the past (e.g., Corr, Pickering, & Gray, 1995b;
Pickering, Diaz, & Gray, 1995). However, a major contention of this chapter is
that the work that has been carried out to date, plus our own recent work
described in this chapter, has served largely to confirm the complexities that
lurk beneath the deceptive simplicity of RST.

2. WHAT EXACTLY IS REINFORCEMENT SENSITIVITY THEORY (RST)?

RST is a very specific theory. We now restate the basic position here:
Individual variations along one fundamental personality dimension (which could be
called impulsivity) are associated with the degree to which certain effects are elicited
by classically conditioned stimuli that predict reward or relief from punishment.
These effects comprise a characteristic pattern of response tendencies (behavioral
activation and approach) accompanied by an increase in arousal resulting in the
energisation of any resultant behavioral output. These effects are discharged by a
basic brain system (the behavioral activation system; BAS). Interindividual variation
in the functional capacity of the BAS is the biological substrate of impulsivity.
Individual variations along a different personality dimension (which could be called
anxiety) are associated with the degree to which certain effects are elicited by
classically conditioned stimuli that predict punishment or frustrative nonreward, plus
novel or innately fear-provoking stimuli. These effects comprise a characteristic
pattern of response tendencies (behavioral inhibition) accompanied by an increase in
arousal resulting in the energization of any resultant behavioral output. These effects
are discharged by a basic brain system (the behavioral inhibition system; BIS).
Interindividual variation in the functional capacity of the BIS is the biological
substrate of anxiety. Impulsivity and anxiety, so defined, are assumed to be
orthogonal personality dimensions because the underlying variations in functional
capacity of the BIS and BAS are assumed to be independent of one another. The key
personality traits of impulsivity and anxiety can be measured using appropriate self-
report scales.
The link between RST and Eysenck's theory was originally forged by stating
the combinations of scores on Eysenck's Extraversion and Neuroticism scales
which may measure the personality dimensions fundamental to RST. In
Eysenckian personality space, trait anxiety was conceived as running from the
neurotic introvert (high anxiety) to stable extravert (low anxiety) quadrants;
impulsivity from the neurotic extravert (high impulsivity) to stable introvert
(low impulsivity) quadrants. It was rapidly realized that these proposals were
 Individual differences in reactions to reinforcing stimuli 39

too simplistic: a good case can be made that Eysenck's Neuroticism is closer to
trait anxiety than to impulsivity, and similarly EPQ-Psychoticism may be of
more relevance to impulsivity than Extraversion (Diaz & Pickering, 1993;
Rocklin & Revelle, 1981).
The basic position, just stated, has received only minor embellishments since
it was first proposed (Gray, 1970), although there are now several scales which
purport to measure BIS and BAS related personality dimensions directly
(Carver & White, 1994; MacAndrew & Steele, 1991; Torrubia & Tobena,
1984). For experiments with human subjects, it is important to note that BIS
and BAS activation may affect performance via the accompanying increase in
arousal. This increase in arousal may improve attention and information
processing. Later, we shall see an example of this in our interpretation of data
from a procedural learning task. One potentially major extension to RST was
the expansion to three dimensions, in parallel with the expansion of Eysenck's
theory (which involved the addition of Psychoticism). The third basic brain
system, linked to the BIS and the BAS, is the fight-flight system. It would thus
be natural to argue for a third fundamental personality dimension, which arises
through individual differences in the functioning of the fight-flight system.
Partly by analogy with Eysenck, therefore, Psychoticism was tentatively
suggested (Gray, 1972, 1987) as a measure of this third personality dimen-
sion. The specific proposals for the neurobiology of the BIS, BAS, and the
fight-flight system are not the main focus of the current chapter; they have
been covered in detail elsewhere (Gray, 1994, 1995).
While the specificity of RST might seem admirable in terms of allowing the
theory to be refuted, it may have contributed to some of the difficulties in the
body of evidence which has been collected. For example, if a piece of research
did not include all the relevant components of the specific theory, then it may
fail to constitute an adequate test of the theory. On the other hand, if an
adequate test appears to reject the theory, then it remains possible that a more
general theoretical statement linking personality and reinforcement may still
have some validity; it could simply be that one or more specific elements of
RST were wrong while the others were correct. Any of the following may have
been wrongly specified: the basic nature of the personality dimensions and
their characterizing labels; the associated patterns of behavioral output; the
requirement that the behavioral responses occur largely to conditioned (rather
than unconditioned) stimuli; the grouping of the conditioned eliciting stimuli
into their separate classes (appetitive vs. aversive); the orthogonal relationship
between anxiety and impulsivity; and finally the specific scales suggested to
assess the basic personality dimensions. So as not to prejudge some of these
aspects of the theory, in this chapter we will generally refer to the key
personality dimensions by discussing "individuals with a reactive BAS (or
40 Personality

BIS)," rather than by referring to impulsive or anxious subjects respectively. As

will become clear, the trait measures which we have found to be correlated
with the effects of reinforcement vary considerably from study to study.
With the 20/20 vision of hindsight, one might question the plausibility of
some aspects of RST (as stated above), purely on logical grounds. The above
statement deliberately referred to the "functional capacity" of the brain
systems rather than their actual functioning. Although the functional capacity
of BIS and BAS might, as assumed, be largely independent, the functioning of
the two systems (in terms of their net behavioral effects) may not. In this way, a
subject with a high score on a personality trait reflecting BIS reactivity may
respond to stimuli that activate the BIS to an extent which also reflects their
(independent) individual level of BAS reactivity. Some form of BIS-BAS
interaction may seem inevitable, given that previous graphical representations
of the systems (e.g., Gray, 1975) have depicted reciprocal connections between
the brain systems, usually in the form of mutual inhibition. In addition, this
interaction takes place prior to the action of the BIS and BAS on behavioral
output; and because the systems have partially opposing effects (inhibition vs.
activation, respectively) on behavioral output, there is a further opportunity for
the two systems to interact.
While it is possible that one system (the system more strongly engaged by
the prevailing eliciting stimuli) could gain total control of behavior, it seems
more plausible to suggest that the net behavioral effect of the activity of the
two systems will reflect a more complex mixture of their joint activation. We
have carried out recent modeling studies of possible BIS-BAS interactions,
drawing on the wealth of knowledge in the neural network literature on
competitive interactions (e.g., Grossberg, 1976). These studies show that when
BIS and BAS reactivity vary independently, very many patterns of behavioral
outcome are possible, whenever BIS output inhibits candidate responses and
BAS output activates them. These modeling results may help us understand
the variety of experimental findings reported in this chapter.
Many experiments in the literature have set out to test the version of RST set
out above. We contend that the multiple components to the theory, however,
have meant that few (if any) published experiments contain all the elements of
this specific formulation. We would argue that many (if not all) experiments,
including some of our own, have failed to meet the necessary behavioral
conditions which must be obtained in order to test the theory fully. The
empirical results obtained, however, may still inform one's view of less specific
versions of RST. We feel it is therefore critical to develop a clear set of
behavioral requirements for testing RST. The critical conditions for experi-
ments are clarified in the following section with the use of a hypothetical
idealized experiment. We intend that this section should allow the reader to
 Individual differences in reactions to reinforcing stimuli 41

evaluate published work on RST against this standard, and to consider

critically the data from our own laboratory, to be described later in this
chapter.

2.1 How may RST be tested empirically?

Our idealized experiment tests the predictions of RST in relation to the BAS.
(One can construct an analogous experiment for testing RST in relation to the
BIS.) Imagine a subject performing a computerized operant task for a reward.
(In studies with human subjects, the "rewards" employed are usually financial
incentives; these are clearly not primary reinforcers but highly overlearned
secondary reinforcers. Although RST reserves the term reward for primary
reinforcers, we shall continue to talk about rewards in the human experimental
context.) The subject is trained until an asymptotic level of performance is
reached, measured for example by response times. In the test of RST a
previously neutral stimulus (such as a tone) will be converted into a con-
ditioned stimulus (CS) for reward via classical conditioning processes. The
effects of the CS on the baseline operant task will then be measured (the
change in response time occurring during phases of the task when the CS
occurs vs. phases without the CS occurring). Therefore, to measure the effect
specifically attributable to the conditioned nature of the CS, its effect before
conditioning (if any) should be estimated. It is useful (see later) to use two
different stimuli; for tones one might be high, and one low, in pitch. The
baseline testing might show that the stimuli have small equivalent effects on
performance (such as a slowing of response caused by a transient distraction
effect).
In a subsequent conditioning phase, one of the tones (randomly selected) is
paired with reward. The reward in this phase is not contingent upon any
operant behavior on the part of the subject; the idea is to establish a classically-
conditioned association between the tone and reward. The subject might be
asked to watch their "winnings total" on the computer screen and observe
whether it increased at any time. Winnings would increase during or shortly
after the sounding of the tone. At the end of this associative phase, the subject
should be tested for learning about this association, for example, by being
asked to predict when their winnings were about to increase. This testing is
necessary to establish that a significant degree of conditioning has occurred
during this phase. Furthermore, one can explore whether there are any
personality effects associated with the classical conditioning phase of the study.
It is important to note that RST makes predictions about the effect of CSs (of
various kinds) on behavior, not about the ease or difficulty with which certain
kinds of CSs may be acquired.
42 Personality

In the final phase of the experiment, the subject should return to the operant
task. Once again the effects of the two stimuli should be assessed. The
untrained tone stimulus should continue to have the same small effects as
previously. By contrast, the tone now conditioned to be a CS for reward should
have a positive motivating effect on performance, improving response speed
across all subjects. This effect should be significantly different from the effects
of the CS before it was paired with reward. Finding a significant motivating
effect across all subjects serves to confirm that the degree of classical
conditioning obtained in the experiment is sufficient to produce motivating
effects on performance. The central prediction of RST is that subjects with a
reactive BAS should show a greater motivational effect of the CS for reward
than BAS-nonreactive counterparts. ("Counterparts" here connotes that,
while the subjects differ on personality measures relevant to the BAS, they
should be matched on trait measures relevant to the BIS. In a between-groups,
median-split approach, the groups should differ on BAS traits and be matched
on BIS traits. In a regression analysis, the correlation of the BAS trait with
behavior should be shown to hold even after partialling out any effects due to
BIS traits.)
In passing, one might ask why it is not simpler to look at the effect on
behavior of an existing CS for reward—money itself. Subjects with a highly
reactive BAS should be predicted to show a bigger speeding up of task
performance when money is delivered to them during task execution than low
BAS subjects. We shall describe examples of this approach later, and it is
commonplace in RST investigations which use financial reinforcers. A
potential difficulty with this design is that one has no experimental control
over the subject's acquisition of the CS; subjects may differ widely in the extent
to which money has acquired "CS-for-reward" properties. Any individual
differences in behavior in the experiment may thus derive from, or be affected
by, differential CS acquisition rather than from the differential capacity of a
CS-for-reward to elicit BAS-based behavior. In the idealized design one checks
the extent of CS acquisition at the end of the conditioning phase; an ideal
evaluation of RST would require that the personality traits of interest are
unrelated to variations in strength of CS acquisition across subjects. In
addition, the strength of the classically conditioned CS reward relationships
acquired by subjects should be partialled out when exploring the personality
correlates of the effects of the CS on the acquired behavior in the final phase of
the design.
We believe that this is the only kind of experiment which directly tests the
basic specific version of RST in relation to the BAS. It is our view that studies
of RST in the literature show widely varying degrees of approximation to this
design; their findings must impact on the specific version of RST to varying
degrees accordingly. It is also worth noting that the idealized experiment
assesses the motivating effects of a CS for reward on an already-established
 Individual differences in reactions to reinforcing stimuli 43

operant behavior. The study does not measure any differential learning for
BAS-active and BAS-inactive subjects. One could, of course, modify the design
to look at a motivating effect on learning. This would involve conditioning an
initially-neutral CS to act as a signal for reward (or for any of the other
reinforcing events relevant to RST). In principle, one could then take a
learning task and explore individual variations in the effect of the CS on the
rate of learning. As a control for the effects of the CS that were not due to the
properties acquired in the conditioning phase, one should also ideally take a
different group of subjects and show that, without the conditioning phase,
there were only small effects of the stimulus on learning, and that these effects
did not covary with BAS- (or BIS-) related personality traits.
Another way to explore learning in relation to RST would be to assess the
reinforcing effects of the newly-acquired CS for reward. The reinforcing effects
of a stimulus are measured by the extent to which subjects are prepared to
acquire a new behavior in order to gain that stimulus; in other words, the
extent to which an operant behavior, upon which the target stimulus is
contingent, can be acquired. This approach would involve a further layer of
complexity on the basic design sketched above, and so we have focused
primarily on the more straightforward motivation tests of RST.
In the above hypothetical example, the use of distinct training, conditioning,
and testing phases is critical. Few studies testing RST in the field of human
personality have made these discrete training phases explicit. In many studies it
appears that the classical conditioning phase is being assumed to have been
achieved (without explicit training) by the use of experimental instructions.
The effects on an already-acquired behavior are also not usually investigated;
the behavior being affected is usually acquired during the course of the test of
the effects of the secondary reinforcers. This may be a critical point as
Pickering et al. (1995) have shown that the personality correlates of maze-
crossing behavior under differing reinforcers were different while the correct
route across the maze was being learned, when compared with the correlates
for performance after the correct route had been acquired. The various studies
on go/no-go discrimination learning are a good example of how the separate
phases of the idealized experiment have been intertwined in a complex
mixture. It is perhaps not surprising that the results from these studies have
been complex, confusing or unexpected (Hagopian & Ollendick, 1994; Zinbarg
& Revelle, 1989). We have recently published one study which began to
approach the idealized experimental design sketched above (Corr et al.,
1995b), and we discuss this later.
We now turn to recent data from our own laboratory. The experiments to be
described all attempted to explore (parts of) the basic version of RST given
earlier, and stand in varying degrees of approximation to the idealized
experiment just outlined. We feel the variety of fit with RST, found across
these studies, is strong evidence that (1) there is a clear link between
44 Personality

reinforcement sensitivity and personality traits but (2) the basic version of RST
needs revision. We are currently preparing a larger paper along these lines; the
current chapter foreshadows some of the issues and in so doing acknowledges
the debt owed to Eysenck both for the work we have carried out to date, and
for the foundations of our future investigations. We shall present our findings
according to the degree of "fit" between the data and RST. These studies are
also summarized in Table 3.1.

2.2 Data with a good fit to the predictions of RST

The term "good fit" implies that the experimental findings accord directly with
the basic version of RST sketched earlier. We describe two studies where we
have obtained an outcome of this kind; the first used the so-called "Q task"
and the second used a procedural learning task.
2.2.1 The Q task. The experimental task was essentially the same as that
reported by Newman, Wallace, and Arnett (1995), and was designed to
measure punishment-induced behavioral inhibition in human subjects. The
task was divided into two phases: conditioning and test. During the
conditioning phase, subjects were presented with a series of letter strings on
a computer screen. Subjects were required to press a response key as fast as
possible in response to each letter string, provided that the string did not
contain the letter Q. Responses to strings containing a Q were punished with a
loss of points. The conditioning phase served to establish the letter Q as an
inhibitory CS associated with punishment, and the extent to which this CS
subsequently elicited behavioral inhibition was assessed during the test phase.
Note the relatively good correspondence with the idealized experiment,
presented earlier.
During the test phase, subjects were presented with a series of displays
containing a mixture of letters and numbers. Subjects were required to press
the response key as quickly as possible in response to each display, provided
that the display did not contain a number. Responses to strings containing a
number were punished with a loss of points. On half of the trials containing
letters only, the letter Q appeared in the display. The presence of this
inhibitory CS resulted in significantly slower response times when compared to
response times to letter-only displays which did not contain a Q. Subjects were
thus showing behavioral inhibition in response to the letter Q.
The basic version of RST sketched earlier predicts that anxious individuals
will show enhanced behavioral inhibition in response to aversive CSs, and
subjects were therefore divided into low and high anxious groups using the
Spielberger State-Trait Anxiety Inventory (Spielberger, Gorsuch, Vagg, &
Jacobs, 1983). Figure 3.1 (Thornton, unpublished data) shows that, consistent
with the predictions of RST, there was a significant interaction between trial
 Individual differences in reactions to reinforcing stimuli 45

Figure 3.1 Effects of state anxiety on behavioral inhibition as measured by the Q task. Error bars
indicate two standard errors.

type (Q-present vs. Q-absent) and anxiety group (low vs. high), reflecting
greater inhibitory effects of the letter Q in high anxious subjects. The figure
presents the data with the subjects split by state anxiety, although the pattern
was essentially identical for analyses using trait anxiety. Reassuringly, these
results replicated Newman et al.'s (1995) findings using the same question-
naire. However, it should be noted that several other good candidate measures
of the relevant BIS trait (anxiety) failed to show the effect shown for the
Spielberger measures (see Table 3.1). Variability of results across question-
naires is a theme which we shall return to below. We are currently using the Q
task to test behavioral inhibition in clinically anxious patients, and in normal
volunteers following administration of the benzodiazepine, diazepam.

2.2.2 The procedural learning task. RST, and Eysenck's related theory of
personality, are characterized by an emphasis upon activation in phylo-
genetically old brain systems underlying the major dimensions of personality,
and by the importance placed on fundamental learning processes. One form of
learning which has been claimed to be phylogenetically old (Reber,
Walkenfeld, & Hernstadt, 1991) is procedural learning; it may also occur
without awareness, although this is a highly contentious claim (see Shanks &
St. John, 1994, for a thorough discussion). This type of learning might
therefore be ideal for testing RST.
In our recent study (Corr, Pickering, & Gray, in press), the procedural
learning task consisted in a long series of reactions to a target that moved
between four locations on a computer monitor. Some movements were
random and others followed specific, predictable patterns. Participants pointed
to the target with a wand which activated a touch-sensitive screen; the target
 ON
Table 3.1. Recent studies from our own laboratories investigating RST

Significant effects by
condition

Study Reinforcement ependent variable(s) HAS questionnaires S questionnaires used Control Punishment Reward
manipulation used

Q task (Thornton, Phase I: Q associated RT to letter-only I 7 (Imp) STAI(Y1,Y2), SP, N/A +STAI(Y1,Y2)@ N/A
unpublished) with punishment. character strings BISQ
Phase 2: Q present or
absent as distractor;
within-subjects.

Procedural learning Control vs. punishment RT difference between I7(lmp) STAI(Y2) -EPQ(P), -STAI(Y2) -EPQ(P), +STA1(Y2) N/A
(Corr et al., in press) (removal of money); random and predictable
between-subjects stimuli
Maze learning Control, reward Maze crossing time I7(lmp, Vent) STAI(Y2), SP Before, None Before, +SP [males], Before, -I7(Vent)
(Pickering et al., 1995) (money) and before and after After, None +EPQ(N) [males], [males].
punishment (removal acquisition -EPQ(E) [males]. After, -I,(Vent)
of money); between- After, +EPQ(N) [males], -I-EPQ(N)
subjects [females] [females], +SP
Two-phase task Phase 1: 3 screen RT; Number of operant TPQ(NS.RD) I7(lmp, TPQ(HA) STAI(Y2) RT: -TPQ(NS) RT: None RT: +STAI(Y2)
(Corr et al., 1995b) colours associated with reinforcers delivered Vent) EPI(Imp) NR N/A NR: +\T(lmp) NR None
neutral, reward and (NR)
punishing USs.
Phase 2: Colours signal
operant reinforcement;
within-subjects
CARROT task (Powell Reward (money) vs. Card- sorting rate Study 1: BB(RR,FS,D) Study 1: BB(BIS) Study 1: +I7(Imp) Study 1: N/A Study 1: +BB(RR)@
et al, in preparation) nonreward conditions; I7(Imp,Vent) *Study 2: TPQ(HA) +SSS(Dis) +SSS(TAS) Study 2: N/A -BB(FS) @ -I7(Imp)
within-subjects SSS(BS,ES, Dis,TAS) +EPQ(E) +BB(BIS) @ -SSS(ES) @
"Study 2: TPQ(NS,RD) Study 2: -TPQ(HA) -SSS(Dis) @
+TPQ(NS) -EPQ(E) @
-BB(BIS) @
Study 2: -TPQ(HA) @
+TPQ(NS) @

Startle Reflex Positive vs. negative vs. EMG amplitude; TPQ(NS,RD) TPQ(HA) These data were analyzed for slide valence (Pos/
Modulation (Corr et al., neutral slides; within- eyeblink RT Neg/Neut) X personality interactions; the
1995a) subjects personality measures for which these interactions
were significant are given below:
Amplitude HA (see Figure 3.3)
Eyeblink RT EPQ(E): only high E Ss showed a
one-way valence effect EPQ(N): only low N Ss
showed a one-way valence effect

Startle Reflex Positive vs. negative vs. EMG amplitude; TPQ(NS,RD) TPQ(HA) See cell above
Modulation (Kumari et neutral film clips; eyeblink RT Amplitude None (see
al., 1996) within-subjects Figure 3.4) Eyeblink RT
EPQ(P): only high P Ss
showed a one-way
valence effect

Geller-Seifter Task Reward (win points) vs. Number of response I7(Imp) STAI(Y1,Y2) SP BISQ N/A None None

1
(Thornton, reward plus punishment sequences and rate of
unpublished) (lose points and loud sequence entry
noise); within-subjects

Note. All experiments used the EPQ except for those marked with an asterisk. RT = response time. USs = unconditioned stimuli. Ss = subjects. EMG
= electromyograph. N/A means the dependent variable/reinforcement condition did not apply, or was not analyzed, in this experiment. +/- indicates
the direction of the significant correlation between personality questionnaires and dependent variables (median split analyses are also treated as
correlations). @ means the correlation in the reinforced condition relates to a difference measure (reinforced condition-control condition) for the
dependent variable given in the table. In the analyses of eyeblink RT data from the startle experiments, "one-way valence effects" principally reflected a
decrease in RT for the negative stimuli compared with neutral.
EPQ = Eysenck Personality Questionnaire; EPI = Eysenck Personality Inventory (Imp = Impulsiveness); TPQ = Cloninger's Tridimensional
Personality Questionnaire (NS = Novelty Seeking; HA = Harm Avoidance; RD Reward Dependence); STAI = Spielberger's State-Trait Anxiety
Inventory (Yl = State Anxiety; Y2 = Trait Anxiety); I7 = (from the) Eysenck Personality Scales (Imp = Impulsiveness; Vent = Venturesomeness); SP
= Torrubia and Tobena's Signals of Punishment Questionnaire; SSS = Zuckerman's Sensation Seeking Scale (ES = Experience Seeking; BS =
Boredom Susceptibility; Dis = Disinhibition; TAS = Thrill and Adventure Seeking); BB = Carver and White's BIS-BAS scale (RR = Reward
Responsiveness; FS = Fun Seeking; D = Drive; BIS = Behavioural Inhibition); BISQ = MacAndrew and Steele's BIS Questionnaire.
48 Personality

then moved to another location; participants then continued to follow the

target as it moved between four locations. As shown by Lewicki, Hill, and Bizot
(1988), there was a selective decline in response times relative to random
targets; this difference reflects procedural learning. Fifty subjects served in a
control condition and 50 in a punishment condition. The signals of
punishments—losses of money from an initial gift—were noncontingent, as
perceived by subjects. All subjects in the punishment condition received similar
numbers of punishment signals throughout the task. The results showed that in
addition to a main effect of (EPQ) Psychoticism (high Psychoticism scores

52

Control
Punishment

47

42

1
"
_S 37
2

32

27

22
-1SD + 1 SD
Trait anxiety
Figure 3.2. Mean procedural learning (ms, ±1 s.e.m.) showing the performance of low and high
trait anxiety (±1 SD) groups in control and punishment conditions.
 Individual differences in reactions to reinforcing stimuli 49

were associated with impaired learning) over both conditions, there was an
interaction of reinforcement condition and scores on Spielberger's Trait
Anxiety scale. Under control conditions, anxiety was negatively correlated with
learning; under punishment, it was positively correlated. This crossover
interaction is shown in Figure 3.2.
These data reveal that high trait anxiety impaired learning under control
conditions but facilitated it under punishment, with the reverse pattern of
effects found for low anxiety. This experiment is an example of our modified
ideal design in which the motivating effects of conditioned reinforcer were
explored, not on an established operant behavior, but on a learning task. RST
predicts that the behavioral inhibition produced by conditioned stimuli
signifying punishment should be greatest for BIS-reactive (high anxiety)
subjects. We did not predict that punishment conditions would facilitate
procedural learning by increasing behavioral inhibition. Instead, we presumed
that the increased arousal levels, indirectly resulting from increased BIS
activity, would improve attention and stimulus processing, thereby facilitating
procedural learning. By making this assumption, the experiment provides clear
support for RST.
Finally, we should consider the finding that high Psychoticism scores were
associated with impaired learning in both control and punishment. This might
be taken as consistent with the evidence (see Zuckerman, 1989) that high
scores on narrow impulsiveness, measured by the EPQ Psychoticism scale, may
be associated with impaired conditioning. Alternatively, if EPQ Psychoticism is
an index of BAS reactivity, as we have argued is plausible on the grounds of the
item content (e.g., Diaz & Pickering, 1993), then this might be evidence of a
phenomenon, to be described shortly, which we term a "complementary-trait
effect."

2.3 Data with an "interesting" fit to the predictions of RST

Interesting effects, in the terminology adopted here, are surprising findings
which can, nevertheless, probably be accommodated post hoc into a suitably
altered version of RST. The degree of interest that such results may stimulate
will probably derive from the plausibility of the post hoc assumptions required
to accommodate the troublesome data, and the degree to which these
assumptions will generate testable ideas.
2.3.1 Combined effects: Response activation and behavioral inhibition. A good
example of an "interesting" effect was reported by us recently (Pickering et al.,
1995). In the learning phase of a maze-crossing task, we found maze-crossing
speeds to be significantly reduced, across all subjects, in the explicitly
reinforced conditions (reward, punishment) compared with the control
condition in which financial incentives were not employed. In the control
50 Personality

condition there were no significant correlations with any of a variety of

measures of anxious and impulsive personality traits. In the punishment
condition, however, trait anxiety variables showed a significant positive
correlation with time to cross the maze: the more anxious subjects were
slower to cross the maze. These results might, at first glance, look like excellent
evidence of behavioral inhibition in highly anxious subjects in a punishing
context. Unfortunately, further analyses revealed that it was the low anxious
subjects who showed increased speed in the punishment condition (relative to
low anxious subjects in the control condition); high anxious subjects showed
little difference between the conditions.
We offered an explanation of these results by considering the inhibitory and
excitatory effects of the punishment condition to be partially dissociable; in
RST, they would both be discharged by BIS outputs. The argument required a
BIS-independent source of excitation; that is, sizable excitation occurring for
even the least BIS-reactive (low anxious) subject in the punishment condition.
Given this excitatory effect, then a superimposed inhibitory effect, reflecting
BIS output and related to the level of trait anxiety, could lead to the observed
results. For low anxious subjects, the anxiety-independent excitatory effect
would be unopposed by any inhibition and the result would be fast maze
crossing in the punishment condition. For high anxious subjects, the inhibitory
effect could approximately cancel out the excitatory effect and lead to only
relatively small changes compared with the control condition. Although
plausible, this account ultimately requires us to specify why the results turned
out as they did in this study while a very different pattern of results was found
in the procedural learning experiment reviewed earlier.

2.3.2 Unexpected effects of sex. Another kind of interesting result occurs when
unexpected variables are found to influence the basic predictions of RST. One
such variable appears to be the sex of the subject. We (Pickering et al., 1995)
found that the effects of personality and reinforcement were influenced by the
subject's sex. In the reward condition, the correlation between maze-crossing
speed and one measure of impulsivity (venturesomeness from the I7; S. B. G.
Eysenck, Pearson, Easting, & Allsop, 1985) revealed that venturesome subjects
crossed the maze faster than nonventuresome subjects; the corresponding
correlation had been nonsignificant in the control condition. While this result
is in accordance with RST, the correlations were significant for male subjects
only, and this finding was true both before and after the correct route across
the maze had been learned. In the punishment condition, there was an even
more striking effect. During maze learning, the correlations between maze-
crossing speed and anxiety (outlined above) derived almost entirely from the
male subjects; after maze learning the significant correlations with anxiety (in
the same direction) were found only for female subjects.
 Individual differences in reactions to reinforcing stimuli 51

We interpreted these results in the light of the debriefing comments of the

participating subjects (who were recruited largely by advertisement in the
general population): males suggested the financial incentives were a very
important motivation for participation; female subjects suggested they were
more intrinsically interested in the experiment. This would be entirely
consistent with the finding that the correlations between performance and
personality in the financially reinforced conditions were largely restricted to
the male subjects. Where significant personality correlations were found for
female subjects (in punishment, after the correct route across the maze had
been learned), very few punishments were administered as performance was
near-perfect for all subjects; in reward, by contrast, financial reinforcers
continued to be administered for perfect performance after the maze route had
been mastered and male subjects continued to reveal correlations with
performance.
We considered that these findings, if replicable, could have major
implications for RST. We have therefore tried to replicate an effect of sex in
a different task. The task we have used is a very simple card-sorting task (which
we refer to as the Card-Arranging Reward Responsiveness Objective Test, or
CARROT for short). The procedure is described in detail by Powell, Al-
Adawi, Morgan, and Greenwood (1996). Subjects have to sort a deck of cards
into one of three piles depending on the presence of a 1, 2, or a 3 in the five-
digit number printed on the card. The subject is familiarized with the task
under baseline conditions and then it is repeated under a nonrewarded
condition in which no explicit positive reinforcement is delivered. The task is
further repeated under reward conditions in which the experimenter gives the
subject 10 pence for every five cards sorted. Although we commented earlier
that this was not the ideal design, the conditioned stimulus for reward,
acquired by all subjects through everyday experience, is taken to be the money
delivered to the subject during the reward condition. The motivating effect of
these reward signals was measured as an increase in the speed of sorting
relative to the nonrewarded condition. Before considering sex effects, it is
interesting to note that, in an open trial of a small group of brain-injured
patients suffering from clinically severe motivational problems ("abulia"),
Powell et al. showed that the dopamine agonist, bromocriptine, significantly
increased the effectiveness of the reward incentive. At baseline (without
bromocriptine), these patients showed an average increase in sorting rate in
reward of less than 2%, whereas at the maximum bromocriptine dose they
showed an average increase in sorting rate in reward of 10%. These findings
paralleled dramatic increases in therapists' independent ratings of the patients'
motivation during therapy sessions, showing that the simple card-sorting task
has ecological validity. In addition, the results are particularly interesting in
relation to the proposed neurological localization of the BAS (see Gray, 1994)
in brain regions richly innervated by dopaminergic neurons.
52 Personality

In recent work (Powell et al., in preparation) we have also looked at the

CARROT in relation to personality traits in two studies with undergraduate
students. From the earlier study (Pickering et al., 1995), we predicted that the
motivating effect in the reward condition would be significantly greater for
male than for female subjects. In both studies this prediction was confirmed.
These sex effects were still present after the effects of personality traits had
been partialled out. As with the earlier study, there was also some suggestion
that the personality correlates of reward motivation differed across males and
females.
2.3.3 Variations in trait measures which correlate with the effects of rein-
forcement. Another interesting finding across studies is the great variability in
the trait measures which relate significantly to behavior under reinforcing
conditions. This can be seen easily in Table 3.1 which presents a summary of all
our experimental work reviewed in this chapter.
One of the studies which illustrated this most clearly was the work on the
CARROT task with undergraduate subjects (Powell et al., in preparation).
The first undergraduate study had large numbers of subjects and used a diverse
battery of questionnaires. We found that the increase in card-sorting rate
under reward showed a weak, but significant, positive correlation with the
Reward Responsiveness subscale of the BIS-BAS questionnaire (Carver &
White, 1994). The BAS scale therefore appears to be validated by these data,
as it was intended to measure the personality trait associated with res-
ponsiveness to reward signals. The study also included a number of other
measures which could plausibly be viewed as "BAS traits." Many of these other
measures correlated more strongly, but negatively, with the increase in card-
sorting rate in the reward condition. The significant relationships across all
subjects were found for: the Fun Seeking subscale of the BIS-BAS question-
naire; the Impulsiveness scale of the I7; and the Experience Seeking and
Disinhibition subscales of the Sensation Seeking Scale (SSS; Zuckerman,
1979). Whilst it was reassuring that a number of different, but similar,
instruments shared the same relationship with the behavioral response to
financial rewards, it was puzzling that this correlation was negative. Part of the
reason appeared to be that these traits were also positively correlated with
card-sorting rate during the baseline (nonrewarded) trial. The highly impulsive
and sensation seeking subjects were sorting the cards very rapidly in the
baseline condition and therefore were able to show only relatively small
increases in speed when explicitly rewarded. While this makes intuitive sense, it
leaves RST to explain why these traits correlated with behavior when no
explicit reinforcement was given.
Multiple regression analyses of these data indicated that ly-Impulsiveness,
the Disinhibition subscale of the SSS, and the Reward Responsiveness subscale
of the BIS-BAS questionnaire, all made separate contributions to the
 Individual differences in reactions to reinforcing stimuli 53

prediction of the increase in card-sorting rate under reward. This may, of

course, suggest that each questionnaire is independently sampling part of the
variation in BAS functioning. Equally plausibly, it may suggest that there are
multiple determinants of a subject's responsiveness to reward signals; if so RST
may be too simplistic in ascribing all this variation to a single brain system. It
should also be pointed out that the other subscales of the questionnaires
involved, despite being equally good candidate "BAS" measures, showed very
small correlations with behavior. This very variable pattern of relationships
across trait measures is, as already noted, found across the whole of Table 3.1.
Understanding this variability must be a major goal for future research on
RST.
2.3.4 Complementary-trait effects. The final category of "interesting" effects
represents a type of finding which we have obtained across several very
different procedures and with different questionnaire instruments. These
findings therefore represent the most consistent result amongst the various
outcomes reported in this chapter. We refer to these results as
"complementary-trait" effects because the personality trait that is found to
correlate with performance in a particular reinforcement condition is the
complementary trait to that predicted to correlate by RST. In particular, we
have repeatedly found that low BIS-reactive subjects show the biggest
behavioral effects in reward reinforcement conditions, rather than the highly
BAS-reactive subjects one would expect to respond according to RST.
A study in which complementary-trait effects occurred was that reported by
Corr et al. (1995b). The study is of particular interest because it approximated
quite closely to the idealized design presented in this chapter. There was an
initial phase of classical conditioning: Subjects learned to associate each of
three colors on a computer screen with the occurrence of a financial penalty
(punishment), financial reward, or no financial reinforcement, respectively.
The subjects' learning of the color-reinforcement associations was measured
during the first phase and was found to correlate significantly with various
personality dimensions relevant to RST. Although the effects on this form of
associative learning were rather weak, the fact that some correlations were
significant is a difficulty for the version of RST sketched earlier. Cloninger's
(1987) Harm Avoidance scale, which can be considered a measure of trait
anxiety, was significantly positively correlated with the subjects' ability to
predict the reinforcing events associated with the aversive color but not those
associated with the other two color stimuli. A similar result (although a
negative correlation) was found for EPQ-E. Cloninger's Reward Dependence
scale showed a significant positive correlation with the amount of learning
shown for associations with the appetitive color. These results seem to imply
that the personality dimensions described by RST may also have an influence
on the strength with which classically conditioned associations are formed. If
54 Personality

an anxious subject can learn more about the stimulus conditions which are
associated with (and predict) aversive events, then RST would require some
fundamental restructuring.
By analogy with the idealized example given earlier, the simplest experiment
which we might have done would have been to explore the effects of the
conditioned stimuli, formed during the first phase of the experiment, on an
established approach behavior. The effect of each of the colored stimuli, after
the classically conditioning phase, could then have been measured and related
to personality measures. In fact we adopted a different design in the
instrumental phase of the experiment (Corr et al., 1995b). In this phase the
classically conditioned screen colors, in a within-subjects design, simply served
as continuous signals that particular reinforcement contingencies (reward,
punishment, no reinforcements) were operative. The performance rules which
determined the delivery of the appropriate reinforcers for these conditions
(see below) were not explained to the subjects; the rules had to be acquired by
trial-and-error using the reinforcing feedback provided. The operant task itself
measured reaction times (RTs) to moving targets, using a computer-screen
layout identical to that used in the procedural learning task described earlier.
When the control color was present on the screen, no financial reinforcers
were delivered; when the punishment color was present, subjects had to learn
to increase (behaviorally inhibit) their mean response times for that period—in
fact they had to be 5% slower than a comparison control period—in order to
passively avoid loss of money; when the reward color was present, subjects had
to learn to decrease (behaviorally activate) their mean response times by 5%
relative to the comparison period in order to gain financial rewards. The
number of reinforcers delivered during the punishment and reward periods
were also recorded; high numbers of punishments thus indicated poor passive
avoidance/behavioral inhibition; high numbers of rewards indicated good
approach behavior/behavioral activation.
We found, during the punishment condition, that high scores on lj-
Impulsiveness were associated with failure to learn the response slowing
required to passively avoid the punishments. This result makes good intuitive
sense: impulsive subjects could not modulate their response speed
appropriately (when slowing of responses was required) and therefore they
received more financial penalties. The findings represent a complementary-
trait effect in that I7-Impulsiveness, which would be expected to relate to the
functioning of the BAS, exerted its effects selectively under conditions
designed to activate the BIS (the administration of punishments unless passive
avoidance occurred).
Another complementary-trait effect was found by Corr et al. (1995b). In the
hypothetically BAS-activating reward condition, low scores on Spielberger's
Trait Anxiety scale were associated with fast responses. This effect was not
present in the neutral condition (signaled by the colored conditioned stimulus
 Individual differences in reactions to reinforcing stimuli 55

which had previously been paired with neither reward nor punishment). The
effect of trait anxiety in the reward condition was largely due to an increased
response speed among low trait anxious subjects rather than a decrease in
response speed for high trait anxious subjects.
We have already commented that the latter finding of Corr et al. (1995b) has
occurred in several of our studies. In addition to the results already reported,
in Pickering et al.'s (1995) maze-crossing study the low anxious subjects were
also faster in the reward condition than comparable subjects in the non-
reinforced control condition. This effect was significant only after the subjects
had fully mastered the task and were no longer making errors in the route
chosen to cross the maze. In Corr et al. (1995b), the correlations with
instrumental learning performance were reported only for the final three
blocks of their task when subjects were at a performance asymptote. Additional
results found with the CARROT task also showed that low trait anxiety
facilitated behavioral activation (the rate of card-sorting). This result was
significant in both of the undergraduate studies reported earlier, and the BIS
effects were shown, by multiple regression, to be separate from the BAS and
sex effects described earlier in this chapter. One should note that, owing to
CARROT'S use with intellectually-impaired clinical samples, this task was
deliberately designed to be virtually error free from the start.
From these studies, one might suggest that low anxiety facilitation of
behavioral activation by reward conditions may perhaps be reliably detected
only when the task has been learned and the subject is striving simply to
improve the speed of performance. Pickering et al. (1995) speculated that high-
level cognitive mechanisms may modulate the effects of performance-related
reinforcement. Hence, an error (and punishment) during the early stages of
learning a response may be expected by the subject and not be especially
motivating in comparison with an error (or the need to avoid one) when the
task has largely been mastered. Such mechanisms would represent an added
level of complexity in the cross-species comparison between psychopharmaco-
logical effects in laboratory animals and effects related to human personality.
We have reported one further demonstration of the same pattern of
complementary-trait effects in the study by Corr et al. (1995a), which looked at
the influence of personality traits on the startle reflex. The startle reflex
consists of a set of involuntary responses to a sudden, intense stimulus,
especially when it is novel and aversive. (Clearly, the argument that cognitive
task-appraisal mechanisms may complicate the effects of reinforcement
manipulations in studies of RST is unlikely to apply to this task.
Nonetheless, the qualitative pattern of results, as we shall see, was similar to
those found in the earlier "operant" experiments.)
In human beings, the most easily measured and most reliable component of
the startle reflex is the magnitude of the eyeblink response to a short, loud
acoustic probe. There has been increased interest in this measure since the
56 Personality

discovery that it can be influenced by prevailing emotional state: unpleasant

emotional states augment eyeblink magnitude, while pleasant hedonic states
reduce it (Lang, Bradley, Cuthbert, & Patrick, 1992). Subjects with a tendency
towards fearfulness and phobia show greater reflex potentiation to unpleasant
mental images relative to pleasant or neutral images (Cook, Hawk, Davis, &
Stevenson, 1991); similar results have been obtained for positively- and
negatively-toned slide stimuli (Greenwald, Bradley, Cuthbert, & Lang, 1991).
Given these results, the startle reflex paradigm, at the very least, seems to offer
a promising tool with which to investigate the brain basis of individual
differences in emotional arousal. To use the startle reflex paradigm to
investigate RST, one must assume that the emotional material used to
modulate the startle reflex has selective effects on the BIS (activated by the
negative stimuli) or the BAS (activated by the positive stimuli). For example, in
using slides as modulating stimuli, this assumption is akin to the assumption
made for money in the financial reinforcement studies. Positive slides are
assumed to include images similar to visual stimuli which have been classically
conditioned (through everyday experiences prior to the experiment) by
appetitive unconditioned stimuli (USs); negative slides are assumed to
contain images that have been classically-conditioned by aversive USs. This
kind of experiment, as with the use of money, sacrifices the experimental
control over the classical conditioning phase that is possible in the idealized
experiment. Furthermore, the startle reflex itself must be assumed to be
sensitive to the outputs of the BIS (with a potentiating effect) and the BAS
(with an attenuating effect). This assumption emphasizes that the outputs of
the BIS and the BAS do not influence operant behaviors solely, although
operant tasks have been the main source of data addressing RST. These
assumptions are supported by the finding that startle potentiation is reduced
by anxiolytic drugs (Lang et al., 1992); these drugs have a selective action on
the behavioral effects of the BIS in animals (Gray, 1977).
In our first study with the startle reflex, Corr et al. (1995a) found that the
Harm Avoidance personality trait (HA; Cloninger, 1987), which is closely akin
to trait anxiety, interacted with the affective tone (positive/negative) of slides.
Only the high HA subjects reacted to unpleasant slides with a potentiated
startle reflex, and only low HA subjects reacted to pleasant slides with an
attenuated startle reflex (Figure 3.3). Personality traits potentially measuring
BAS reactivity did not relate to startle reflex modulation. With the
assumptions made earlier, the highly BIS reactive subjects showed the
greatest startle potentiation by negative emotional material, as predicted by
RST. The recurring complementary-trait pattern was also observed once again:
subjects with low scores on BIS-related traits demonstrated large responses to
signals of reward (the positive slides) rather than subjects scoring highly on
BAS-related personality traits.
 Individual differences in reactions to reinforcing stimuli 57

Figure 3.3. Mean EMG amplitudes (arbitrary units) for pleasant, neutral, and unpleasant slides for
subjects low (HA-) and high (HA+) on (TPQ) Harm Avoidance. Only HA+ subjects showed
significant potentiated startle to unpleasant slides, and only HA- subjects showed significant
attenuated startle to pleasant slides.

The ability to demonstrate the effects of positive reinforcement in low BIS-

reactive subjects in particular makes good sense, post hoc, under RST. Most
experimental scenarios are probably somewhat anxiety provoking; we would
therefore expect the BIS of most subjects to be in active state prior to the
delivery of reinforcers in the experiment. The output of the BIS, as we have
already commented, serves in part to antagonize the output of the BAS. It is
therefore possible that the varying level of BIS output across subjects would
58 Personality

add considerable noise to any correlations between performance and the

varying BAS output stimulated by positive reinforcers in the experiment.
Subjects with a weakly reactive BIS may not be anxious in the experiment and
the effects of BAS output may be more easily observed in such subjects. These
ideas could also explain the tendency to detect the complementary-trait effects
on easy or already mastered tasks: experimental anxiety in low anxious subjects
would be particularly likely to be minimized in these cases. This interpretation
would imply that, when looking for BAS-related personality effects, partialling
out any BIS effects first could reveal BAS-trait correlations that may otherwise
be hidden in reward conditions. The results from the CARROT task show that
independent BIS and BAS related correlations can be observed in reward
conditions, although the greater difficulty, in general, in obtaining BAS-trait
correlations may partly reflect the trait measures used, given the uncertainties
over which trait measures best reflect BAS reactivity (see above).

2.4 Data with a clearly negative fit to the predictions of RST

We have obtained a clearly negative fit in studies where the behavior, across
subjects, is indicative of the action of the BIS and/or BAS, but the behavioral
effects do not covary with any of the relevant personality dimensions. We
describe two studies where we have obtained an outcome of this kind: the first
used a so-called "Geller-Seifter conflict task"; the second set of negative
outcomes were found in our later studies with the startle paradigm.
2.4.1 Geller-Seifter conflict task. A large number of animal studies invest-
igating the psychopharmacology of anxiety have used the Geller-Seifter
conflict test (Geller & Seifter, 1960) to assess the effects of antianxiety drugs
on punished responding. During a typical Geller-Seifter test, rats are first
trained to press a bar in return for food reward. Following the initial training
phase, so-called "conflict" periods are indicated by the presentation of a
distinctive CS such as a tone or light, during which bar-pressing continues to be
rewarded, but is also occasionally punished with footshock. The presence of
the CS results in suppression of bar-pressing (behavioral inhibition), and
consistent with RST, this suppression is reliably attenuated by the admini-
stration of anxiolytic drugs (Cook & Davidson, 1973).
We hypothesized that, if behavioral suppression during conflict periods
reflects activation of the BIS in the rat, it should follow that a similar
suppressive effect in Man would be strongest in anxious individuals. To test
this prediction, an experimental task was developed to provide a human
analogue of the Geller-Seifter conflict test.
During the experimental task, subjects were required to enter number
sequences into a computer in return for winning points which were later
exchanged for money. Subjects were informed that their aim was to discover a
 Individual differences in reactions to reinforcing stimuli 59

winning number "code" which would win points when entered into the
computer, and that a new winning code would be generated each time that an
old code was discovered. The task thus required the subject to enter a variety of
sequences in order to discover as many winning codes as possible during the
task. This use of the entry of number codes as an operant behavior was based on
earlier studies of human learning by Vogel-Sprott (1967). There were in fact no
winning codes, and sequence entries were rewarded with the winning of points
according to a variable ratio (VR) reinforcement schedule. Conflict periods were
signaled by a change in the color of the computer screen, during which a VR
punishment schedule was superimposed onto the reward schedule. Punishments
took the form of a mildly aversive 116 dB white noise delivered via headphones,
plus a message on the computer screen indicating a loss of points.
We (Thornton, unpublished data) have found, in close correspondence with
the animal data, that subjects entered significantly fewer number sequences
during conflict periods when compared to reward-only periods, and also took
significantly longer to enter each individual sequence. The effect of anxiety on
the size of this suppression effect was assessed by dividing subjects into low-
and high-anxious groups, using a variety of different anxiety measures. Against
the prediction of RST, high-anxious subjects showed no sign of increased
behavioral suppression during conflict periods, relative to low-anxious subjects.
While these results are a serious failure to find evidence in support of RST,
there are possible mitigating arguments. First, it may be that the effects of
anxiety predicted under RST are apparent only for very high levels of anxiety
such as are present in clinically anxious individuals. We are therefore currently
using this task to test patients with anxiety disorders. Second, it is possible that
the behavioral suppression observed using this task may be due to some factor
other than the activation of the BIS, in which case no effects of anxiety would
be expected. This issue will be clarified by experiments currently underway to
investigate the effects of the anxiolytic drug, diazepam, on the performance of
this task by healthy volunteers. We may find that task performance is affected
by diazepam in human subjects in the same way that the rats' behavior in the
original Geller-Seifter experiments was affected by anxiolytic drugs. If so, then
the two tasks are likely to involve similar brain systems. As RST was developed
from data suggesting that anxiolytic drugs in the rat have a selective action on
the BIS, then the behavioral suppression in the human task (if affected by
anxiolytic drugs) could more confidently be inferred to reflect BIS reactivity.
While this would suggest that humans possess a BIS-like brain system, the null
effects with respect to anxious personality traits would therefore strongly reject
RST's alignment of these personality traits with functional variation in the
fundamental brain systems discovered in animal research.
60 Personality

2.4.2 Further startle reflex experiments. Kumari et al. (1996) used affectively-
toned short filmclips, rather than slides, as the stimuli acting to modulate the
startle reflex. In this study, no influence of any of the EPQ or TPQ (Cloninger,
1987) variables was found on the degree of startle reflex modulation as
measured by response amplitude or magnitude (see Table 3.1 for an effect of
EPQ-P on response latency). This is a thoroughly negative finding under the
predictions of RST which were sketched out earlier. The null results are all the
more troubling given the "interesting" fit with RST found in our previous work
with startle reflex modulation and personality (Corr et al., 1995a), although
poor replicability of personality correlations is a feature of some other
published findings concerning RST (e.g., Hernaiz-Sanders, Pickering, & Gray,
1994).
A number of explanations seem possible, however, for the failure to
replicate the personality effects found earlier by Corr et al. (1995a). First, one
could question whether we should expect a "replication" of the personality
effects given the substantial change of modulating stimuli between the studies.
However, if the startle paradigm offers a method for studying "emotional"
responses, then its ecological validity might be seriously questioned if it can be
used with only a narrow class of modulating stimuli. Second, the film material
produced stronger modulation than that elicited by slides. In Corr et al.'s
study, only high Harm Avoidance scorers showed startle potentiation by
negative slides, and only low Harm Avoidance scorers showed startle attenu-
ation by positive slides. In Kumari et al. (1996), by contrast, all relevant
personality groups showed bidirectional startle modulation to some degree
(see Figure 3.4). These data might therefore suggest that, under conditions
which evoke more intense anxiety or fear (or other relevant emotions), the BIS
and BAS may be activated in both weakly and highly reactive subjects. This
may mitigate against the detection of significant effects related to individual
differences in trait anxiety or impulsivity. This explanation would, of course,
mean that RST applies only to low-intensity emotional stimuli and would
further imply that the individual differences concerned would not be relevant
in any real-world settings where high-intensity emotional stimuli prevail. This
is not an attractive scenario, as it would raise serious doubts about the
relationship between RST and clinical conditions such as anxiety, which are
usually presumed to relate to high-intensity emotional stimuli.
Third, in human subjects, social anxiety (e.g., arising from a social en-
counter) produces startle attenuation and not potentiation, particularly in
introverts (Blumenthal, Chapman, & Muse, 1995). The content of Eysenck's E
and N scales are mainly concerned with social anxiety. In the social context of
an experiment, therefore, high N/low E subjects, although perhaps not showing
startle attenuation to aversive stimuli, might show a very small degree of startle
potentiation; this result is the opposite to that which one would predict under
RST. There is a suggestion of weak potentiation by high N/low E subjects in
 Individual differences in reactions to reinforcing stimuli 61

Figure 3.4. Mean eyeblink response amplitudes (scores after logarithmic transformation; error
bars display ± s.e.m.) for pleasant (POS), neutral (NEU), and unpleasant (NEG) filmclips for the
low and high Extraversion (E- and E+) and Neuroticism (N- and N+) groups.

the data obtained by Kumari et al. (see Figure 3.4). Furthermore, in Corr et al.
(1995a), only high E and low N groups showed startle potentiation as
measured by response latency (see Table 3.1). Accepting this argument would
imply a fractionation of anxiety itself and might therefore limit the application
of RST to certain anxiety subtypes only. Once again, this is an unattractive
argument, because RST was intended to form part of a general account of
anxiety.
62 Personality

Finally, startle potentiation in human subjects may not be a valid tool for
investigating RST. Lang and his associates have advocated the paradigm as a
means for measuring an individual's general "negative emotionality" (Lang et
al., 1992). As such, it could be argued that RST, which relates to individual
differences in the elicitation of the specific negative emotion of fear/anxiety,
would best be tested with specific negative stimuli. This has clearly not been
the case in the startle studies we have conducted. This argument is even more
serious in light of the evidence that negative stimuli eliciting distinct emotions
(e.g., fear and disgust) do not potentiate the startle reflex comparably (Balabun
& Taussig, 1994). The effects of positive stimuli may be equally complex. As a
general rule, however, one should be very wary of denouncing an experimental
procedure as inappropriate in a post hoc fashion, when the results are not as
one originally predicted. If one cannot agree whether a theory has been
properly tested in a particular study then the theory needs restating even more
clearly. This is one of the primary goals of this chapter.

3. CONCLUDING COMMENTS

It was our intention in this chapter to illustrate the diversity of findings that
emerge from experimental tests of RST. It has not been difficult to provide
such illustrations, despite the fact that they have all been chosen from our own
research. In the light of this diversity, we clearly cannot reach any kind of
optimistic conclusion about the degree to which the theory is supported by the
available data, much as we would like to be in a position to do so. Yet the
literature is full of single experimental studies which end on just such an
optimistic note, with apparent justification in the data they report. What gives
rise to this difference? In part, it stems from our having pursued an extended
research program in which we have repeatedly tried to apply the same
theoretical principles, but through the prism of different experimental
paradigms. If we had conducted just one or two experiments, it would have
been disarmingly easy, post hoc, to interpret (as distinct from predict) a variety
of possible findings in the light of RST. We fear that a large number of the
findings in the literature that apparently support RST do so only in this very
weak sense. When, however, one surveys the results from a large number of
different studies using different paradigms, but all based on the same
theoretical principles, it becomes clear that the post hoc interpretations that
work for the results of one do not apply to others in the set. It is disturbing,
furthermore, that, even when the results of different studies do provide direct
experimental support for predictions derived from RST (either a priori or post
hoc), the particular personality questionnaire that turns up trumps itself then
differs from study to study (see Table 3.1).
 Individual differences in reactions to reinforcing stimuli 63

Faced with this bewildering array of outcomes, it might be tempting to

abandon RST altogether. Yet one cannot but remain impressed by the sheer
frequency with which significant relationships nonetheless do emerge between
one or other relevant personality trait and one or other relevant change in
behavior due to reinforcement effects. Somewhere in the human brain there
clearly are systems which influence individual differences in sensitivity to
reinforcement (as there certainly are in the rodent brain; Gray, 1987); and in
some way or other the activity of these systems relates to existing questionnaire
measures of personality, whether these predate RST or were more recently
devised in the light of this theory. Furthermore, despite the overwhelming
success of Eysenck's biological theory of personality in accounting for
individual differences in other domains, his theory does not appear to be
able to encompass the kind of personality-reinforcement interactions that we
have summarized here and which have also been reported by others
(Derryberry & Reed, 1994; Newman, 1987; Zinbarg & Revelle, 1989;
Zuckerman, 1991). So, if not RST, then son-of-RST will be needed, either
as an extension to Eysenck's theory or as a separate but complementary
edifice.
We cannot at present discern where the changes in existing RST should best
be made. Tasks which, a priori, have apparently equally good credentials as
tests of the theory (e.g., the Q task and the modulated startle reflex described
above) have provided very different degrees of support for the theory.
Similarly, personality measures which have equally good theoretical lineages,
and indeed correlate quite highly one with another, show quite diverse patterns
of predictive accuracy for behavior in different tasks. Our best advice for the
moment therefore (and we plan to follow it ourselves) is to continue to devise a
range of tasks, each with as sound a theoretical basis as possible, and to use
these to extend the database until the underlying consistencies—which we are
confident must exist—eventually become clear. Some reliable patterns (the
complementary-trait and sex effects) have started to emerge in the data we
have just presented. We may also look with confidence to important new help
from emerging biological technologies, such as neuroimaging (e.g., N.S.Gray,
Pickering, & J.A.Gray, 1994) or molecular and quantitative genetics (e.g.,
Benjamin, Li, Patterson, Greenberg, Murphy, & Hamer, 1996; Ebstein et al.,
1996; Flint et al., 1995). Final understanding may require that we trace the
route in detail from the molecular genetic bases of individual differences, via
the neuroanatomy and neurochemistry of the brain systems that these genetic
factors influence, to the eventual behavioral interactions with a complex and
changing environment for which these brain systems are responsible. But this,
of course, is exactly what Eysenck's own massive contribution to the field has
always led us to expect.
64 Personality

REFERENCES

Balabun, M. T., & Taussig, H. N. (1994). Salience of fear/threat in the affective

modulation of the human startle reflex Biological Psychology, 38, 117-131.
Benjamin, J., Li, L., Patterson, C, Greenberg, B. D., Murphy, D. L., & Hamer, D. H.
(1996). Population and familial association between the D4 dopamine receptor gene
and measures of Novelty Seeking. Nature Genetics, 12, 81-84.
Blumenthal, T. D., Chapman, J. G., & Muse, K. B. (1995). Effects of social anxiety,
attention, and Extraversion on the acoustic startle eyeblink response. Personality and
Individual Differences, 19, 797-807.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and
affective responses to impending reward and punishment: The BIS/BAS scales.
Journal of Personality and Social Psychology, 67, 319-333.
Cattell, R. B. (1957). Personality and motivation structure and measurement. Yonkers-on-
Hudson, New York: World Book Co.
Cloninger, C. R. (1987). A systematic method for clinical description and classification
of personality variants: A proposal. Archives of General Psychiatry, 44, 573-588.
Cloninger, C. R., Svrakic, D. M., & Przybecky, T. R. (1993). A psychobiological model
of temperament and character. Archives of General Psychiatry, 50, 975-990.
Cook, E. W. Ill, Hawk, L. W., Davis, T. L., & Stevenson, V. E. (1991). Affective
individual differences and startle reflex modulation. Journal of Abnormal Psychology,
100, 5-13.
Cook, L., & Davidson, A. B. (1973). Effects of behaviourally active drugs in a conflict-
punishment procedure in rats. In S. Garattini, E. Mussini, & L. O. Randall (Eds.),
The benzodiazepines (pp. 327-345). New York: Raven Press.
Corr, P. J., Wilson, G. D., Fotiadou, M., Kumari, V., Gray, N. S., Checkley, S., & Gray,
J. A. (1995a). Personality and affective modulation of the startle reflex. Personality
and Individual Differences, 19, 543-553.
Corr, P. J., Pickering, A. D., & Gray, J. A. (1995b). Personality and reinforcement in
associative and instrumental learning. Personality and Individual Differences, 19, 47-
71.
Coir, P. J., Pickering, A. D., & Gray, J. A. (in press). Personality, punishment and
procedural learning: A test of J. A. Gray's anxiety theory. Journal of Personality and
Social Psychology.
Costa, P. T., & McCrae. R. R. (1985). The NEO personality inventory manual. Odessa,
FL: Psychological Assessment Resources.
Derryberry, D., & Reed, M. A. (1994). Temperament and attention: Orienting toward
and away from positive and negative signals. Journal of Personality and Social
Psychology, 66, 1128-1139.
Diaz, A., & Pickering, A. D. (1993). The relationship between Gray's and Eysenck's
personality spaces. Personality and Individual Differences, 15, 297-305.
Ebstein, R. P., Novick, O., Umansky, R., Priel, B., Osher, Y., Elaine, D., Bennett, E. R.,
Nemanov, L., Katz, M., & Belmaker, R. H. (1996). Dopamine D4 receptor (D4DR)
exon III polymorphism associated with the human personality trait of Novelty
Seeking. Nature Genetics, 12, 78-80.
Eysenck, H. J. (1953). The structure of human personality. London: Methuen.
 Individual differences in reactions to reinforcing stimuli 65

Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Charles

Thomas.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality
Questionnaire (Adults). London: Hodder & Stoughton.
Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for
impulsiveness, venturesomeness and empathy in adults. Personality and Individual
Differences, 6, 613-619.
Flint, J., Corley, R., DeFries, J. C., Fulker, D. W., Gray, J. A., Miller, S., & Collins, A.
C. (1995). A simple genetic basis for a complex psychological trait in laboratory mice.
Science, 269, 1432-1435.
Geller, G., & Seifter, J. (1960). The effects of meprobamate, barbiturates, d-
amphetamine and promazine on experimentally-induced conflict in the rat.
Psychopharmacologia, 1, 482-492.
Gray, J. A. (Ed.) (1964). Pavlov's typology. London: Pergamon.
Gray, J. A. (1970). The psychophysiological basis of introversion-extraversion. Behavior
Research and Therapy, 8, 249-266.
Gray, J. A. (1972). Causal theories of personality and how to test them. In J. R. Royce
(Ed.), Multivariate analyses and psychological theory (pp. 409-463). London:
Academic Press.
Gray, J. A. (1975). Elements of a two-process theory of learning. London: Academic
Press.
Gray, J. A. (1977). Drug effects on fear and frustration: Possible limbic site of action of
minor tranquillizers. In L. L. Iverson, S. D. Iverson, & S. H. Snyder (Eds.), Handbook
of psychopharmacology: Vol. 8. London: Wiley.
Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.),
A model for personality (pp. 246-276). Berlin: Springer.
Gray, J. A. (1987). The neuropsychology of emotion and personality. In S. M. Stahl, S.
D. Iversen, & E. C. Goodman (Eds.), Cognitive neurochemistry (pp. 171-190).
Oxford: Oxford University Press.
Gray, J. A. (1994). Framework for a taxonomy of psychiatric disorder. In S. H. M. van
Goozen, N. E. Van de Poll, & J. A. Sargeant (Eds.), Emotions: Essays on emotion
theory (pp. 29-59). Hove, U.K.: Lawrence Erlbaum Associates.
Gray, J. A. (1995). A model of the limbic system and basal ganglia: Applications to
anxiety and schizophrenia. In M. S. Gazzaniga (Ed.), The cognitive neurosciences (pp.
1165-1176). London: Bradford Books.
Gray, N. S., Pickering, A. D., & Gray, J. A. (1994). Psychoticism and dopamine D2
binding in the basal ganglia using single photon emission tomography. Personality and
Individual Differences, 17, 431-434.
Greenwald, M. K., Bradley, M. M., Cuthbert, B. N., & Lang, P. J. (1991). Fear and
loathing: Startle reflex and visceral response. Psychophysiology, 28, 526.
Grossberg, S. (1976). Adaptive pattern classification and universal receding: Part I.
Parallel development and coding of neural feature detectors. Biological Cybernetics,
23, 121-134.
Guilford, J. P., & Zimmerman, W. S. (1956) Fourteen dimensional temperament
factors. Psychological Monographs, 70 (10), 1-26.
66 Personality

Hagopian, L. P, & OUendick, T. H. (1994). Behavioral inhibition and test anxiety: An

empirical investigation of Gray's theory. Personality and Individual Differences, 16,
597-604.
Hernaiz-Sanders, H., Pickering, A. D., & Gray, J. A. (1994). Is response latency an
optimal variable to test Gray's model of personality? In B. DeRaad, W. K. B.
Hofstee, & G. L. Van Heck (Eds.), Personality psychology in Europe: Vol 5 (pp. 162-
167). Tilburg, The Netherlands: Tilbury University Press.
Kumari, V., Corr, P. J., Wilson, G. D., Kaviani, H., Thornton, J. C, Checkley. S. A., &
Gray, J. A. (1996). Personality and affective modulation of the startle reflex by
emotionally-toned filmclips. Personality and Individual Differences, 21, 1029-1041.
Lang, P. J., Bradley, M. M., Cuthbert, B. N., & Patrick, C. J. (1992). Emotion and
psychopathology: A startle proble analysis. In L. Chapman, & D. Fowles (Eds.),
Progress in experimental personality and psychopathology research. New York:
Springer.
Lewicki, P., Hill, T., & Bizot, E. (1988). Acquisition of procedural knowledge about a
pattern of stimuli that cannot be articulated. Cognitive Psychology, 20, 24—37.
MacAndrew, C., & Steele, T. (1991). Gray's behavioural inhibition system: A
psychometric examination. Personality and Individual Differences, 12, 157-171.
Newman, J. P. (1987). Reaction to punishment in extraverts and psychopaths:
Implications for the impulsive behaviour of disinhibited individuals. Journal of
Research in Personality, 21, 464—480.
Newman, J. P., Wallace, J. F., & Arnett, P. Behavioral inhibition system functioning in
anxious, impulsive and psychopathic individuals. Manuscript submitted for publication.
Pickering, A. D., Diaz, A., & Gray, J. A., (1995). Personality and reinforcement: An
exploration using a maze-learning task. Personality and Individual Differences, 18,
541-558.
Powell, J. H., Al-Adawi, S., Morgan, J., & Greenwood, R. J. (1996). Motivational
deficits after brain injury: Effects of bromocriptine in 11 patients. Journal of
Neurology, Neurosurgery, and Psychiatry, 60, 416-421.
Powell, J. H., Lessiter, J., Bridges-Palmer, J., & Pickering, A. D. (1997). Individual
differences in behavioural activation: Measuring and modelling the responses to
signals of reward. Manuscript in preparation.
Reber, A. S., Walkenfeld, F. F., & Hernstadt, R. (1991). Implicit and explicit learning:
Individual differences and IQ. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 17, 888-896.
Rocklin, T., & Revelle, W. (1981). The measurement of extraversion: A comparison of
the Eysenck Personality Inventory and the Eysenck Personality Questionnaire. British
Journal of Social Psychology, 20, 279-281.
Shanks, D. R., & St. John, M. F. (1994). Characteristics of dissociable human learning
systems. Behavioral and Brain Sciences, 17, 367-447.
Spielberger, C. D., Gorsuch, R. L., Lushene, R., Vagg, P. R., & Jacobs, G. A. (1983).
Manual for the State-Trait Anxiety Inventory: STAI (form Y). Palo Alto, CA:
Consulting Psychologist Press.
Torrubia, R., & Tobena, A. (1984). A scale for the assessment of "susceptibility to
punishment" as a measure of anxiety: Preliminary results. Personality and Individual
Differences, 5, 371-375.
 Individual differences in reactions to reinforcing stimuli 67

Vogel-Sprott, M. D. (1967). Partial-reward training for resistance to punishment and to

subsequent extinction. Journal of Experimental Psychology, 75, 138-140.
Zinbarg, R., & Revelle, W. (1989). Personality and conditioning: A test of four models.
Journal of Personality and Social Psychology, 57, 301-314.
Zuckerman, M. (1979). Sensation-seeking: Beyond the optimal level of arousal. Hillsdale,
NJ: Erlbaum.
Zuckerman, M. (1989). Personality in the third dimension: A psychobiological
approach. Personality and Individual Differences, 10, 391-418.
Zuckerman, M. (1991). Psychobiology of personality. Cambridge: Cambridge University
Press.
Chapter 4
Temperament and personality: Eysenck's three
superfactors as related to temperamental
dimensions

/. Strelau and B. Zawadzki

1. INTRODUCTION

The history of personality research, which goes back to ancient times, clearly
shows that from the very beginning two basic approaches developed in
grasping the nature of personality. One of them, which has its roots in the
Hippocrates-Galen typology of temperament, consisted of explaining human
behavior by referring to the individual's organism, that is, by means of endo-
genous factors, whereas representatives of the other approach tried to explain
human behavioral characteristics by means of external conditions. The
monograph Characters by Theophrastus (4th/3rd century B.C.), who explained
individual differences in character mainly in terms of environmental settings
(so-called exogenous factors), exemplifies the latter view. The endogenous
approach was essential for the development of trait-oriented theories of
personality including temperament. Although many researchers may be
mentioned who contributed during our century to the development of trait-
oriented personality theories, three of them—G. W. Allport, R. B. Cattell and
H. J. Eysenck—should be regarded as the giants who were most influential in
this field of research (Pervin, 1993). It is H. J. Eysenck to whom this chapter is
not only dedicated but also whose views on personality and temperament
constitute the essence of our considerations and studies presented here.
2. THE TEMPERAMENTAL ROOTS OF EYSENCK'S THEORY OF
EXTRA VERSION AND NEUROTICISM

Eysenck (1970; H. J. Eysenck & M. W. Eysenck, 1985), going back to the

ancient times, emphasizes that the dimensions of extraversion and neuroticism
were anticipated by Hippocrates and Galen, and in most of his publications

The research reported in this chapter was supported by Grant KBN 1H01F0660G.
 Temperament and personality: Eysenck's three superfactors 69

Eysenck notes the similarity between these two basic temperamental dimen-
sions and the Hippocrates-Galen typology.
The causal approach to personality which is most typical for Eysenck was
ascribed by him to Gross (1902) who gave a neurophysiological interpretation
of the primary-secondary function which became one of the three tempera-
mental dimensions (including also activity and emotionality) in Heymans'
(1908) typology. To wind up this historical introduction, the research con-
ducted by Heymans was regarded by Eysenck (1992a) as the first model in
which the psychometric approach was combined with laboratory tests—a
typical paradigm developed in Eysenck's laboratory for studying personality.
To continue the historical perspective regarding the temperamental roots of
Eysenck's personality theory, Pavlov has to be mentioned here. His ideas and
conceptualizations are present in Eysenck's theory of extraversion since the
very beginning of its existence. In the "inhibition theory" of extraversion,
Eysenck (1957) referred among others to Pavlov's (1951-1952) typological
theory of excitation and inhibition when formulating the basic postulates which
guided the physiological and behavioral study of extraversion for at least the
next 10 years.
Pavlov was influential in the development of Eysenck's theory of extra-
version also for at least two other reasons. The classical conditioned reflex
paradigm developed by Pavlov (1951-1952), which was used for assessing the
basic properties of the central nervous system (CNS), has been applied by
Eysenck (1957) to study the nature of personality. Taking as point of departure
this concept, Eysenck hypothesized that due to the greater susceptibility of
extraverts to inhibition they would condition less efficiently as compared with
introverts. The relationship: "conditioned reflex-extraversion" did not change
when Eysenck (1967) introduced the arousal theory of extraversion except for
the fact that the explanation was made not in terms of excitation and inhibition
but by using the construct of arousal. Since speed of conditioning is a positive
function of intensity of stimuli, and arousal is chronically higher in introverts,
conditioning is more efficient in them compared with extraverts.
Eysenck's (1957,1970) drug postulate also has its roots in Pavlov's studies on
the type of nervous systems regarded as the physiological bases of temperament
(Strelau, in press-a). This postulate says, among other things, that stimulant
drugs decrease inhibition and increase excitation, thereby producing introverted
patterns of behavior. One of Pavlov's (1951-1952) standard methods for
diagnosing strength of excitation consisted of increasing pharmacologically the
excitability of the CNS by means of different doses of caffeine. Caffeine
increases the excitability of the nerve cells, thus causing stimuli of weak intensity
to produce similar effects to stronger stimuli under normal conditions.
The selective review of the roots of Eysenck's theory of personality taken from
a temperamental perspective constitutes a suitable background for relating the
Eysenckian Three Superfactors—Psychoticism (P), Extraversion (E), and
70 Personality

Neuroticism (N) to other temperamental conceptualizations developed during

the last two decades of our century. Before doing this, some information is
needed regarding the nature of the Eysenckian superfactors PEN as well as his
understanding of the concepts of personality and temperament.

3. THE NATURE OF PEN: DO THESE PERSONALITY FACTORS BELONG TO

TEMPERAMENT?

If we consider Eysenck's study of personality from a historical perspective, it

was extraversion and neuroticism which became the subject of research from
the 1940s (Eysenck, 1947, 1952, 1957). Early in the 1950s Eysenck (1952)
suggested that psychoticism might be regarded as a third dimension of
personality, but this idea was fully elaborated only two decades later. These
two stages—with psychoticism either absent and present in Eysenck's theory—
are also reflected in the development of instruments aimed at measuring the
Eysenckian superfactors. The Maudsley Personality Inventory (MPI) which
stems from the 1950s (Eysenck, 1956), and the Eysenck Personality Inventory
(EPI, H. J. Eysenck & S. B. G. Eysenck, 1964), which was constructed almost a
decade later, were both aimed at measuring only two superfactors—extra-
version and neuroticism. Psychoticism, seen as a dimension with a status
equivalent to extraversion and neuroticism, was introduced by Eysenck in the
1970s (H. J. Eysenck & S. B. G. Eysenck, 1975).

3.1 The description of PEN

The aim of this chapter is to illustrate relationships between Eysenck's three
superfactors and other constructs to be acknowledged in research on temperament,
and this requires a short description of PEN. The three superfactors "are defined
in terms of observed intercorrelations between traits" (Eysenck, 1990, p. 244).
Extraversion, as opposed to introversion, is composed of such traits as
sociability, liveliness, activity, assertiveness and sensation-seeking. Neuroti-
cism, for which emotionality is used as the synonym, has the following com-
ponents: anxiety, depression, guilt feelings, low self-esteem, and tension. The
opposite pole of neuroticism is emotional stability. Psychoticism, the opposite
of which is impulse control, consists of such primary traits as aggression,
coldness, egocentrism, impersonality, and impulsiveness (H. J. Eysenck &
M. W. Eysenck, 1985).
The three basic factors, which are orthogonal to each other, have the status
of second-order factors. They have a hierarchical structure and are composed
of first-order factors (primary traits) which, in turn, result from a group of
correlated behavioral acts or action tendencies. A closer view of the three
 Temperament and personality: Eysenck's three superfactors 71

superfactors of personality shows that the status or nature of psychoticism is

different from those of extraversion and neuroticism. This can be demon-
strated from at least two points of view.
Eysenck's theory of personality is biologically oriented, and there is ample
evidence, collected by himself and many others, to confirm the biological
nature of PEN (e.g., H. J. Eysenck, 1970; Eysenck & M. W. Eysenck, 1985;
Gale, 1983; Stelmack, 1990). But, whereas the genetic determination of an
essential part of the phenotypic variance of all three superfactors is now
beyond any doubt (e.g., Eaves, Eysenck, & Martin, 1989; Loehlin, 1992), the
physiological evidence for psychoticism is lacking. Also at the theoretical level
Eysenck postulated the existence of a neurological basis for extraversion and
neuroticism (Eysenck, 1967, 1970, Eysenck & M. W. Eysenck, 1985), whereas
he did not develop a physiological theory of psychoticism. Although some
speculations have been made (see Claridge, 1987; Claridge, Robinson, &
Birchall, 1985; Eysenck, 1992b) with respect to the question about the physio-
logical or biochemical basis (correlates) of psychoticism, until now there is no
answer, or at best there are no unequivocal solutions regarding this issue.
Eysenck considered the relationship between personality and various forms of
psychiatric disorders in terms of quantitative differences. Thus it is natural to
assume that neurotics have greatly elevated neuroticism scores and psychotics
should have inflated psychoticism scores (H. J. Eysenck & M. W. Eysenck, 1985).
However, the link between personality and pathology was not so evident in the
case of extraversion. It was mostly the two extreme poles of the extraversion
dimension in combination with extreme scores on the neuroticism dimension by
means of which psychiatric disorders were explained (Eysenck, 1970). The
psychoticism dimension also has a special status among the three factors when
the links between personality and psychiatric disorders are examined. There is an
almost direct relationship between a high score of psychoticism and personality
disorders and, at the very extreme, psychosis. When under environmental stress,
the probability that psychosis occurs is, according to Eysenck (1992b), a
monotonic function of the psychoticism dimension. The link between psycho-
ticism and pathology is most explicitly expressed when Eysenck describes the
extreme poles of this dimension. Low P is characterized by such behaviors as
altruistic, socialized, empathic, conventional, and conformistic. By the way, these
personality descriptions suggest that they belong to the nontemperamental
domain of personality. In turn, high P is characterized by such behavior disorders
and psychotic characteristics as criminal, impulsive, hostile, aggressive, psycho-
pathic, schizoid, unipolar depressive, affective disorder, schizoaffective, and
schizophrenic (Eysenck, 1992b).

3.2 Do the PEN constructs belong to personality or temperament?

The question arises: how are the PEN superfactors of personality related to
temperament? Eysenck's view, unanimously presented in the 1980s (H. J.
72 Personality

Eysenck & M. W. Eysenck, 1985) is, that: "Personality, as we look at it, has two
major aspects: temperament and intelligence" (p. VII). Hence, "temperament,
that is, the noncognitive aspects of personality" (p.353) should be regarded,
according to Eysenck, as a synonym for personality, assuming that intelligence,
and other cognitive characteristics, are not taken into account. This explains
why Eysenck uses in many of his publications both concepts, personality and
temperament, interchangeable. Such a position was also explicitly expressed by
Eysenck's most eminent student, J. Gray (1991).
Taking the position, according to which personality is equivalent to tem-
perament, as a starting point for comparing the Eysenckian superfactors with
other temperamental traits, we are met with some obstacles. There exists a
large number of personality characteristics, such as traits referring to the self-
concept, to social behavior, and to motivation, which many eminent trait-
oriented personality researchers (Allport, 1937; Buss & Finn, 1987; Cattell,
1965; GuUford, 1959) do not classify as belonging to temperament but to other
domains of personality.
The discussion regarding the relationship between the constructs of
temperament and personality is partly an academic dispute (see Strelau, in
press-b), and depending on the understanding of both constructs under
discussion, their relationship may be diverse. However, in spite of differences
in the comprehension of temperament, most researchers will probably agree
with a definition according to which temperament refers to basic, relatively
stable personality traits which are present since early childhood, occur in man,
and have their counterpart in animals.
As has been shown in many studies (Eysenck, 1970, 1990; H. J. Eysenck &
M. W. Eysenck, 1985) the Eysenckian superfactors, but especially extraversion
and neuroticism, fulfill the criteria of the above formulated understanding of
temperament. Thus, whether we take the position of Eysenck, according to
whom personality is equivalent to temperament, or the view represented by
most temperament researchers, according to whom temperament constitutes
only a part of personality, the Eysenckian superfactors should be regarded as
having a temperamental nature.

3.3 The relationship between PEN and other temperamental

dimensions
When comparing the Eysenckian three superfactors with temperament
characteristics which stem from other theories or conceptualizations, the
following facts must be taken into account:
1. PEN dimensions have a biological nature, and the biological background
of two of these factors—extraversion and neuroticism—refers to different
 Temperament and personality: Eysenck's three superfactors 73

physiological and biochemical mechanisms taking part in regulation of the

level of activation (extraversion) and arousal (neuroticism) (H. J. Eysenck
& M. W. Eysenck, 1985);
2. Eysenck's theory of PEN is adult-oriented. The three major tempera-
mental dimensions have been identified and studied in adults, and only
secondarily transferred to children, but in Eysenck's laboratory subjects
were never below school age. With respect to extraversion and neuroticsm,
studies have also been conducted on animals.
These facts explain to some extent, why most of the comparisons in which
the Eysenckian superfactors were related to temperamental characteristics
refer mainly to arousal-oriented traits such as sensation seeking, impulsivity,
anxiety, emotionality, activity, or the Pavlovian CNS properties (e.g., H. J.
Eysenck & M. W. Eysenck, 1985; Strelau, 1983; Strelau & Eysenck, 1987;
Zuckerman, 1991).
They also explain why, for a period of about 30 years, no search for links has
taken place between the Eysenckian PEN and the many temperamental traits
proposed by child-oriented researchers. The latter were centered mainly
around Thomas and Chess, the founders of contemporary research on
temperament in the United States. Only after Eysenck gained popularity
among temperament researchers in the United States (e.g., Bates & Wachs,
1994; Buss & Plomin, 1984; Kagan, 1994) and in Europe, the child-oriented
temperamental constructs have been adapted (e.g., Strelau & Angleitner,
1991), and the search for the links under discussion became intensified (see
Prior, Crook, Stripp, Power, & Joseph, 1986; Ruch, 1994; Windle, 1989;
Zawadzki, 1995).
To provide a more comprehensive view on the relationship between the
Eysenckian superfactors and temperamental traits being representative of
different approaches and assessed by a diversity of inventories, the PEN factors
will be related to traits which stem from conceptualizations on temperament
described below.

3.4 Buss and Plomin's behavior-genetic theory of temperament

According to Buss and Plomin (1984), temperament refers to inherited
personality traits already present in children. The structure of temperament is
composed of traits which fulfill the criteria mentioned in the definition—
genetic determination and presence since childhood. They are the following
traits: sociability, activity, and emotionality. Emotionality has a hierarchical
structure and on a lower level is represented by distress, fear, and anger.
Developmentally speaking, in infancy only distress is present and with age
emotionality becomes more complex with anger as the latest occurring
74 Personality

component. The traits postulated by the authors are present not only in early
childhood but also in adults. According to Buss and Plomin (1984), sociability
is a trait which has much in common with Eysenck's extraversion, and
emotionality (but only the distress and fear components) reminds of
neuroticism.

3.5 Thomas and Chess' interactional theory of temperament

Thomas and Chess' (1977) theory, which has a descriptive status, developed
under the influence of longitudinal studies aimed at examining the role of
interaction between temperamental traits and environment in human develop-
ment and behavior disorders from early childhood onward. According to
Thomas and Chess, the structure of temperament across the life-span is
composed of such characteristics as: activity level, rhythmicity, approach-with-
drawal, adaptability, threshold of responsiveness, intensity of reaction, quality
of mood, distractibility, attention span, and persistence. No relationships with
the Eysenckian superfactors were hypothesized by the authors. In our study the
structure of temperament, as proposed by Thomas and Chess, will be related to
the Eysenckian factors in a slightly modified version as proposed by Windle
and Lerner (1986) who developed an inventory, which allows for the study of
these traits not only in children and adolescents but also in adults.

3.6 The biological theory of sensation seeking developed by Zuckerman

From the very beginning, the theory of sensation seeking was based on the
concept of optimal level of arousal. In Zuckerman's (1979) first theory the
cortical-reticular formation loop, regarded by Eysenck as the biological
substrate for extraversion, constituted the physiological basis of sensation
seeking. After a revision, Zuckerman (1994) postulated a psychopharmacolo-
gical model to explain the biological mechanism determining individual
differences in sensation seeking. The latter construct has a hierarchical
structure and is composed of thrill and adventure seeking, experience seeking,
disinhibition, and boredom susceptibility. Zuckerman's theory is based on
studies conducted on adults. The hypothesized relationships between the
Eysenckian superfactors and Zuckerman's sensation seeking can be character-
ized by two stages. Before psychoticism was incorporated in Eysenck's theory it
was extraversion which was related to sensation seeking. Later it is
psychoticism which, according to Zuckerman (1994), has the closest links
with sensation seeking. It was also predicted that there is no relationship
between neuroticism and sensation seeking.
 Temperament and personality: Eysenck's three superfactors 75

3.7 The Pavlovian properties of the central nervous system

Pavlov's (1951-1952) theory of CNS properties, the composition of which
constitutes the so-called type of nervous system, has gained increasing popu-
larity over the past two decades, especially among biologically oriented
personality (temperament) researchers (e.g., Eysenck, 1972; Gray, 1964;
Strelau, 1983; Zuckerman, 1979). The reason for this renewed interest in the
properties of the CNS may be explained by at least two facts. First, Pavlov's
typology offers the most adequate physiological interpretation of the Hippocr-
ates-Galen types of temperament. Second, the Pavlovian constructs of strength
of excitation and protective inhibition are closely related to concepts of arousal
(activation) to which most biologically oriented personality theories refer (see
Strelau & Eysenck, 1987). The three basic CNS properties are the following:
strength of excitation, strength of inhibition, and mobility of nervous processes.
In this chapter reference will be made to the Pavlovian constructs opera-
tionalized in behavioral terms as proposed by Strelau (1983) and Strelau,
Angleitner, Bantelmann, & Ruch (1990a). Since the 1940s, Eysenck (1947) has
put forward hypotheses regarding the relationships between extraversion and
neuroticism and the Pavlovian constructs. Most explicitly they have been
formulated in Eysenck's (1972) paper directly devoted to comparing these two
superfactors with properties of the nervous system. According to him
"extraverts possess a strong nervous system, introverts a weak one, and ...
the behavioral differences observed between these two 'types' are causally
related to underlying differences in strength of the nervous system" (Eysenck,
1972, p. 176). With respect to neuroticism, Eysenck, referring to Gray's (1964)
consideration regarding the relationship between E and N and the Pavlovian
constructs, hypothesized that the weak nervous system "might have its closest
analogue in the personality field ... with N, and somewhat less highly with
introversion" (Eysenck, 1972, p. 177). When comparing the CNS properties
with PEN, such unequivocal hypotheses have not been put forward by Eysenck
in respect to mobility of nervous processes and to strength of inhibition,
understood as acquired (conditioned) inhibition. Additionally, no relationship
was predicted by him with respect to psychoticism. On the basis of many
studies in which former Eysenckian measures of extraversion and neuroticism
were applied, such as the MPI and EPI, together with the Strelau Tempera-
ment Inventory (STI; Strelau, 1983; Strelau, Angleitner, & Ruch 1990b), which
is a precursor of a PTS version, additional predictions were made. It was
hypothesized (Ruch, Angleitner, & Strelau, 1991) that extraversion correlates
with mobility of CNS processes and that neuroticism correlates negatively with
all three Pavlovian CNS properties.
76 Personality

3.8 Rusalov's theory of temperament based on a functional systems

approach
Rusalov (1989a), who was a student of Teplov and Nebylitsyn, developed his
own theory of temperament which has its roots in the Pavlovian approach,
especially in the neo-Pavlovian conceptualization developed by Teplov (1964)
and Nebylitsyn (1972), and in the theory of functional systems developed by
Anokhin (1978). According to Rusalov (1989a, b), the structure of tempera-
ment is composed of four basic traits such as: ergonicity (closely related to
strength of excitation), plasticity (remaining mobility of CNS processes), tempo
(counterpart to the neo-Pavlovian construct of dynamism of nervous
processes), and emotionality. Guided by the idea, strongly incorporated in
Russian psychology, that man's activity may be directed towards objects
(things) or towards people (social world), Rusalov (1989a) extended the
structure of temperament by separating two facets of each of the four
temperamental traits what resulted in separating eight traits—four object-
related and four socially related. Although Rusalov (1989a), when comparing
the two Eysenckian superfactors—E and N—with his temperamental
constructs, did not make specific predictions regarding the possible relation-
ships, he hypothesized that the socially related traits separated by him will
show the closest links with Eysenck's personality constructs, which has been
partially confirmed in his study.

3.9 The regulative theory of temperament developed by Strelau

The regulative theory of temperament (RTT) developed by Strelau (1983,
1996) in the 1970s has manifold roots such as: Pavlov's (1951-1952)
conceptualization regarding CNS properties, personality theories referring to
the concept of arousal (Eysenck, 1970; Gray, 1964), and the theory of action as
presented in the 1960s by Tomaszewski (1978). The RTT ascribes tempera-
ment to formal characteristics of behavior which are present since early infancy
in humans and have their counterpart in animals. The theory underlines the
biological background of these characteristics subject to slow changes due to
biologically determined life-span variation and individual-specific genotype-
environment interaction. In recent studies (Strelau & Zawadzki, 1993), it has
been demonstrated that the structure of temperament, as viewed from the
RTT, consists of the following six traits: briskness, perseveration, sensory
sensitivity, emotional reactivity, endurance, and activity. Strelau and Zawadzki
(1995) hypothesized that extraversion will be correlated with briskness,
endurance, and activity, and neuroticism—with perseveration and emotional
reactivity. No prediction was made with respect to psychoticism as related to
the RTT traits.
 Temperament and personality: Eysenck's three superfactors 77

4. PSYCHOMETRIC EVIDENCE REGARDING THE RELATIONSHIPS BETWEEN

THE PEN FACTORS AND OTHER TEMPERAMENTAL CONSTRUCTS

Although studies have already been conducted, in which the separate

temperamental traits were related to the Eysenckian superfactors, the data
presented here are to some extent unique. They present the most
comprehensive picture, showing the manifold relationships between the PEN
and the whole variety of temperament constructs, viewed in Section 3.
The presentation of data consists of three stages. The Eysenckian dimen-
sions were assessed by means of the Eysenck Personality Questionnaire—
Revised (EPQ-R; S. B. G. Eysenck, H. J. Eysenck, & Barrett, 1985). There-
fore, some basic characteristics of the Polish adaptation of the EPQ-R will be
given. It is important to show that this inventory fulfills all basic criteria
postulated by the authors. Subsequently, we will present correlational data
which illustrate how the separate temperamental traits are related to
extraversion, neuroticism, and psychoticism as measured by EPQ-R. Finally,
several factor analyses will be performed to show the location of PEN among
the many temperament traits being studied.

4.1 The Polish EPQ-R version

For the purpose of our research we have made use of the Eysenck Personality
Questionnaire—Revised, as adapted to a Polish population by Drwal and
Brzozowski (Drwal, 1995). This inventory has been used in several studies and
the most detailed psychometric characteristics of this instrument are presented
by Zawadzki (1995).
The sample to which our study refers consists of 1.817 subjects aged 21-77
years (mean = 26.8, SD = 12.95). Among them are 903 females (mean = 26.9,
SD = 13.20) and 914 males (mean = 26.7, SD = 12.76). They represent 30
different job categories and their education varies from elementary to
university level.
The EPQ-R was administered on different occasions, often combined with
several other temperamental or personality inventories, and always by means
of personal (direct or indirect) contact with subjects. The basic psychometric
data of the Polish version of EPQ-R are depicted in Table 4.1.
As can be seen, EPQ-R has satisfactory psychometric characteristics which
differ only slightly from the original version (Eysenck et al., 1985). The
differences consist mainly of lower, although acceptable, reliability scores
(alpha coefficient) for the Psychoticism (P) scale in the Polish version. There is
also an absence of gender differences with respect to the Extraversion (E)
scale, whereas, with respect to the original EPQ-R, females score higher than
males. In both versions, males have significantly higher scores on the P scale,
78 Personality

Table 4.1. Basic psychometric characteristics of the Polish version of EPQ-R

Endorse- Cronbach
Scale M SD ment Kurtosis Skewness alpha
P
Total 9.29 3.99 0.29 0.88 0.77 0.70
Females 8.76 3.83 0.27 0.46 0.66 0.69
Males 9.81 4.07 0.31 1.09 0.85 0.69
E
Total 14.47 5.13 0.63 -0.59 -0.48 0.86
Females 14.49 5.17 0.63 -0.42 -0.58 0.86
Males 14.44 5.09 0.63 -0.76 -0.37 0.85
N
Total 12.91 5.72 0.54 -0.95 -0.14 0.87
Females 14.22 5.43 0.59 -0.71 -0.37 0.86
Males 11.60 5.70 0.48 -0.95 0.10 0.87
L
Total 9.11 4.26 0.43 -0.49 0.24 0.80
Females 9.36 4.33 0.45 -0.57 0.13 0.80
Males 8.86 4.18 0.42 -0.37 0.34 0.79
Note. Abbreviations of scales: P = Psychoticism, E = Extraversion, N = Neuroticism, L = Lie
scale. Differences between males and females: E scale (t = 0.22, n.s.), N scale (t = 9.96, p< .01),
P scale (t = 5.62, p < .01), L scale (t = 2.51, p < .02).

and females on the Neuroticism (N) scale. Drwal (1995), who compared five
EPQ-R studies conducted in the U.K. with his Polish data, stated that the small
differences between the original EPQ-R and the Polish version do not exceed
the variations present when data from the original EPQ-R are compared with
each other.
As expected by the Eysencks (Barrett & S. B. G. Eysenck; 1984; H. J.
Eysenck & M. W. Eysenck 1985), the PEN factors, assumed to be universal,
are either orthogonal to each other or they correlate only slightly in EPQ-R
studies. Table 4.2 presents correlations between the PEN scales, including age
and gender.
The structure of temperament, described by the PEN factors, is as
postulated by Eysenck. The coefficient of correlation, although significant
between the E scale and the N and P scales (which is due to the large size of
the sample), are low and do not exceed the score of 0.20 (with minus sign for
the EN correlation). There is no correlation between the N and P scales. These
statements are valid for the whole sample, and separately for males and
females. The coefficients of correlation between the EPQ-R scales are
compatible with Eysenck's data, except for the fact that extraversion and
neuroticism correlates in our study between -0.18 and -0.20 for the total
sample as well as for both sexes, whereas in the study by Eysenck et al. (1985)
the coefficients are below 0.10. The pattern of intercorrelations in our study is,
however, similar to those recorded in several other national versions of the
 Temperament and personality: Eysenck's three super/actors 79

Table 4.2. Intercorrelations between Polish EPQ-R scales, age and gender included

Scale E N L Age Sex

T F M T F M T F M

P .15* .16* .16* -.03 .02 -.02 -.36* -.37* -.34* -.17* .13*
E -.19* -.20* -.18* -.14* -.12* -.16* -.16* -.01
N -.16* -.18* -.18* .02 -.23*
L .28* -.06

Note. Abbreviations of scales as in Table 4.1. T = total sample, F = females, M = males.

Correlations significant at p < .01 (two-tailed) are marked by an asterisk.

EPQ-R (e.g., Drwal, 1995; Eysenck & Haapasalo, 1989; Hanin, S. B. G.

Eysenck, H. J. Eysenck, & Barret, 1991). To conclude, our data confirm the
assumption regarding the orthogonality between the PEN factors, and they are
very satisfactory when compared with other cross-cultural studies.
The pattern of intercorrelations between the EPQ-R scales and gender
replicates a most common finding that neuroticism and psychoticism correlate
with sex, neuroticism being higher in females and psychoticism higher in males,
although these correlations are rather low. In contradiction to several other
studies (e.g., Ruch, 1994; Zuckerman, 1991), no correlation occurred between
gender and extraversion. With respect to age, our data show that there is a
slightly negative link between this demographic characteristic and two of the
EPQ-R scales—psychoticism and extraversion. The same finding came out
when males and females were analyzed separately.

4.2 The PEN factors related to other temperamental traits: correlational

data
Parallel to the six conceptualizations in the temperament domain, six different
inventories, representative of these approaches, have been applied in our study
together with the EPQ-R. Buss and Plomin's behavior-genetic theory of
temperament is represented by the EAS Temperament Survey (EAS-TS; Buss
& Plomin, 1984), the Thomas and Chess' interaction theory of temperament by
the Revised Dimensions of Temperament Survey (DOTS-R; Windle & Lerner,
1986), Zuckerman's sensation-seeking theory by the Sensation Seeking Scale
Form V (SSS-V; Zuckerman, 1979), the Pavlovian approach by the Pavlovian
Temperament Survey (PTS; Strelau & Angleitner, 1994), Rusalov's theory of
temperament by the Structure of Temperament Questionnaire (STQ; Rusalov,
1989a), and Strelau's regulative theory of temperament by the Formal
Characteristics of Behavior-Temperament Inventory (FCB-TI; Strelau &
Zawadzki, 1993). The separate scales of these inventories are listed in Table
4.3 which presents the correlational data.
80 Personality

Depending on which of the inventories was compared with the EPQ-R

scales, the number of subjects varied from N — 174 (for the EPQ-R—STQ
data) to N = 1.689 (the EPQ-R-PTS comparisons). All subjects were recruited
from the sample (N = 1.817) which served for assessing the basic psychometric
characteristics of the EPQ-R.
There is no space to discuss the details regarding the data illustrating the
relationships between the separate scales representative of the different
approaches and the EPQ-R measures. There are, however, several statements
which are quite interesting and on which we will concentrate.
First, the data are almost without exception in accordance with the
hypotheses advanced in the theoretical part of the paper. Thus, it might be
concluded that extraversion correlates, as expected, with: sociability, strength
of excitation, mobility of CNS processes, social-related traits (SEr, SP1, and
STe), briskness, and activity. Neuroticism shows correlations with distress, fear,
all three Pavlovian CNS properties (negatively), perseveration, and emotional
reactivity, and it does not correlate or shows only slight links with the
sensation-seeking scales. Psychoticism is positively related to the sensation-
seeking scales. The only result which goes against the prediction (put forward
by Buss & Plomin, 1984) is that neuroticism correlates with anger.
Second, many data obtained in this study, although not explicitly predicted
by the authors of the various inventories, are in accordance with findings
already presented in the literature. Thus, our data confirm that extraversion
correlates with EAS-TS—activity (e.g., Oniszczenko, 1995), with DOTS-R—
activity-general, approach-withdrawal, and mood quality (e.g., Windle, 1989),
with the sensation-seeking scales (e.g., Eysenck & Haapasalo, 1989; Ruch et
al., 1991), and with STQ—plasticity (e.g., Brebner & Stough, 1993). It must be
added that the consistent correlation between extraversion and activity, as
measured by different scales, is in agreement with Eysenck's (H. J. Eysenck &
M. W. Eysenck, 1985) theory, according to which activity is one of the
components of extraversion. With respect to neuroticism, the following tem-
peramental traits correlate in accordance with data reported in the literature:
DOTS-R—rigidity (e.g., Windle, 1989), STQ—emotionality object-related and
social-related (e.g., Brebner & Stough, 1993; Rusalov, 1989a), and FCB-TI—
endurance and briskness, both with negative sign (Strelau & Zawadzki, 1995).
Third, if we look at coefficients of correlation higher than .30, depicted in
Table 4.3, it is easy to conclude that almost all temperamental scales correlate
with one of the two superfactors—extraversion and neuroticism (or with both).
This finding points strongly to the temperamental nature of extraversion and
neuroticism. With respect to psychoticism, only the sensation-seeking scales
are correlated with this superfactor. None of the correlations representing the
32 remaining temperamental scales reached the level of .30. This result
provocatively raises the question of whether psychoticism is a trait which
belongs to the domain of temperament. Psychoticism is characterized, among
 Temperament and personality: Eysenck's three super-factors 81

Table 43. Pearson correlations between PEN and other temperamental scales

Temperament
scale E N P L
EAS-TS
Sociability (Soc) .51* -.14* .00 .02
Activity (Act) .39* .01 -.00 .01
Fear(F) -.23* .56* -.09* -.02
Distress (D) -.25* .66* -.01 -.09*
Anger (A) .04 .47* .10* -.18*
DOTS-R
Activity level-general (A-G) .33* -.04 .03 .00
Activity level-sleep (A-S) .04 .13* .03 -.12*
Approach-withdrawal (A-W) .48* -.25* .05 -.02
Flexibility-rigidity (F-R) .29* -.33* .07 -.09*
Mood quality (MQ) .44* -.21* .02 -.10*
Rhythmicity-sleep (R-S) -.10* -.01 -.21* .26*
Rhythmicity-eating (R-E) -.01 -.10* -.09* .12*
Rhythmiciry-daily habits (R-H) -.03 .03 -.19* .18*
Distractibiliry (Dist) .07 -.13* -.02 .11*
Persistence (Pers) .01 -.09* -.16* .21*
SSS-V
Total .43* -.18* .51* -.30*
Thrill and adventure seeking (TAS) .29* -.23* .25* -.08
Experience seeking (ES) .27* -.09 .36* -.23*
Disinhibition (Dis) .35* -.08 .43* -.32*
Boredom susceptibility (BS) .29* -.07 .42* -.25*
PTS
Strength of excitation (SE) .40* -.46* .19* .05
Strength of inhibition (SI) -.10* -.32* -.20* .30*
Mobility of CNS processes (MO) .48* -.34* .09* -.01
STQ
Object-related ergoniciry (Er) .26* -.03 -.14 .24*
Social ergonicity (SEr) .81* -.16 -.04 -.06
Temperament scale E N P L
Object-related plasticity (PI) .46* -.17 .13 -.03
Social plasticity (SP1) .38* .36* .28* -.26*
Object-related tempo (Te) .56* -.06 .10 -.06
Social tempo (STe) .52* -.04 .09 -.06
Object-related emotionality (Em) -.13 .70* -.02 -.03
Social emotionality (SEm) -.18 .82* .04 -.12
FCB-TI
Briskness (BR) .30* -.38* .01 .10*
Perseveration (PE) -.10* .59* -.21* -.12*
Sensory sensitivity (SS) .02 .01 -.16* .02
Emotional reactivity (ER) -.34* .71* -.16* -.12*
Endurance (EN) .21* -.49* .11* .11
Activity (AC) .72* -.19* .23* -.19*
Note. Correlations significant at/? < .01 (one-tailed) are marked by an asterisk. The characteristics
of samples to which the separate inventories were administered are as follows: EAS-TS-N = 895,
434 females (F) and 461 males (M), age: range 16-77, M = 31.2, SD = 15.56; DOTS-R-N = 900,
437 F and 463 M, age: range 16-77, M = 31,2, SD = 15.57; SSS-V-N = 534, 324 F and 210 M, age:
range 15-69, M = 20.6, SD = 6.79; PTS-N = 1689, 846 F and 843 M, age: range 15-77, M = 27.1,
SD = 13.40; STQ-N = 174, 72 F and 102 M, age: range 15-70, M = 28.7, SD = 10.58; FCB-TI-N
= 1422, 759 F and 663 M, age: range 15-77, M = 27.2, SD = 13.99.
82 Personality

other things, by such components as: egocentrism, impersonality, antisocial,

and criminal behaviors at one pole, and altruistic, conventional, and conformist
behaviors at the other pole of this dimension (Eysenck, 1992b). If we take the
position, represented by Allport (1937), who distinguished, as different from
temperament, a subdivision of personality known as character, the behavioral
characteristics of psychoticism suggest that this superfactor has much in
common with character. According to Allport, character can be identified with
the ethically effective organization of the individual's forces. The impulsivity
component transferred by Eysenck (H. J. Eysenck & M. W. Eysenck, 1985)
from extraversion to psychoticism seems to have a temperamental nature. If we
consider character, together with Allport, as a disposition to inhibit impulses in
accordance with a regulative principle, the remaining characteristics of psycho-
ticism should be classified as belonging to this subdivision of personality. Such
components of psychoticism, as for example altruism, antisocial, and criminal
behavior, which can be evaluated in a society as being good or bad, develop in
later stages of ontogenesis in contrast to E and N. The links between psycho-
ticism and sensation seeking can be explained basically by the fact that
sensation seeking refers to a given extent to antisocial, illegal behaviors
(Zuckerman, 1994). In various studies it was shown that sensation seeking, in
distinction to psychoticism, correlates with several temperamental character-
istics (Ruch et al., 1991; Strelau & Zawadzki, 1995), which is probably due to
the stimulative component of this trait and refers mainly to activity.

4.3 The location of PEN among temperamental traits

To show the location of the Eysenckian superfactors among other tempera-
mental characteristics, several factor analyses will be presented with different
configurations of temperamental scales combined with the EPQ-R scores. This
procedure allows us to examine the stability of using PEN as markers in these
analyses. In all comparisons, a solution constrained to three factors, based on
principal components with Varimax rotation, was applied. Data from three
samples are presented to provide some insight in the dynamics of PEN,
depending on the number and quality of scales subject to factor solution.
Subjects hi all three samples were high-school and university students.
Sample A consisted of 317 subjects (155 females and 162 males) with mean age
of M = 19.1 (SD = 2.40), sample B comprised 250 subjects, 106 females and
144 males (age: M = 18.5, SD = 3.19), and sample C was composed of 251
students (114 females and 137 males) with age: M = 18.5, SD = 3.10. The
samples did neither differ with respect to the proportion of males and females
(Chi2 = 2.40, df = 2, p = .30), nor did the age differences reach the level of
0.01 significance, and there was an acceptable homogeneity of variance across
samples (Levene test F = 2.66, dfl = 2, df2 = 815, p = .07).
 Temperament and personality: Eysenck's three superfactors 83

Subjects in sample A completed EPQ-R, PTS, FCB-TI, and SSS-V, in

sample B, EPQ-R, EAS-TS, and DOTS-R, and in sample C, EPQ-R, PTS,
FCB-TI, EAS-TS, and DOTS-R. Additionally, in samples B and C the NEO
Five-Factor Inventory (NEO-FFI, Costa & McCrae, 1992) was administered.
In accordance with findings reported in the literature (see Costa & McCrae,
1992; Goldberg & Rosolack, 1994; Strelau & Zawadzki, 1996; Zawadzki,
1995), it was expected that the Extraversion (E) and Neuroticism (N) scales
from the NEO-FFI will show similar loadings on the same factors as both the
corresponding EPQ-R scales. Moreover, according to Eysenck (1994), two of
the Big Five factors—agreeableness (A) and conscientiousness (C), are sup-
posed to be components of psychoticism, and openness (O) is a factor which
belongs to the intellectual domain. As argued by Strelau and Zawadzki (1996;
see also John, 1990), the A and C factors belong rather to the domain of
character than temperament. Perhaps this explains to some extent their close
links to psychoticism.
To examine whether the PEN factors may be regarded as stable markers in
the three-factor solution applied in our study, for each of the three samples
several models of factor analysis will be introduced. They include different
compositions of scales, with the EPQ-R Lie scale present or absent. In
presenting the data, we will concentrate in the discussion on the location of
PEN in the factor solutions without going into details about the structure of the
separate factors.
Table 4.4, based on data from sample A, presents four models of factor
solution, which differ from each other in the number of scales included in the
factor analysis. Model 1, with the smallest number of variables, comprises the
PEN, PTS and FCB-TI scales, Model 2—additionally the Lie scale, in Model 3
the SSS-V scales are added (without the L scale), and Model 4 with the Lie
scale included.
As can be seen, in each of the four models all three PEN scales are present
with loadings 0.70 or above. When the Lie scale was added to the three-factor
solution, it showed the highest loading on the factor identified as P, but did not
influence the factor solution. The inclusion of the SSS-V scales caused only
minor changes in the factor solution. There is a change in factor order, and
additionally, a slight increase occurred in the loading of the P scale (from 0.70
to 0.75-0.76), and a decrease in the loading of the E scale (from 0.88 and 0.86
to 0.76 and 0.77). The loading of the N scale (0.85) was stable across all four
models of factor analysis. It may be concluded that the presence or absence of
the SSS scales in the factor solution did not change the nature of the three
factors which might be identified as PEN.
Factor analysis based on data obtained from sample B comprised the four
following factor solutions: Model 1—EPQ-R (without L), EAS-TS, and
DOTS-R scales, Model 2—the same scales with L included, Model 3—NEO-
FFI scales added without L, and Model 4—all scales included.
84 Personality

Table 4.4. Factor analysis based on EPQ-R, PTS, FCB-TI and SSS-V scales

Model
Scale 1: PEN+T 2: PEN+T+L 3: PEN+T+S 4: PEN+T+L+S
I II III I II HI I II III I II III

P
.11 .14 -.70 .20 .08 -.70 .12 .76 -.19 .17 .75 -.15
E .04 .88 -.11 .04 M -.14 .10 32 .75 .09 .29 .77
N -.85 -.02 .07 -.85 -.05 -.03 -.85 -.05 .04 -.85 -.01 .01
L .28 -.12 .66 30 -.57 .06
SE .64 JO .02 .62 .52 .03 .66 .16 .44 .65 .12 .46
SI .54 -.18 .50 .46 -.11 .55 .50 -39 .03 .48 -.43 .03
MO 33 .67 .15 .29 .70 .10 36 .06 .67 33 .03 .67
BR .55 .51 .20 .50 .54 .22 56 .00 J2 J5 -.05 J3
PE -.79 -.02 32 -.83 -.02 .19 -.80 -.24 .15 -.82 -.20 .11
SS -.11 .19 .69 -.22 .25 .57 -.16 -.22 .43 -.18 -.26 .43
ER -.82 -.30 .09 -.82 -.31 -.00 -.84 -.14 -.20 -.83 -.09 -.23
EN .75 .23 -.00 .74 .25 .05 .77 .03 .18 .76 .00 .20
AC .07 .88 -.09 .07 .87 -.11 .12 .41 .76 .13 35 .79
TAS .33 .46 .27 35 .42 30
ES .04 .57 .23 .07 34 .27
Dis .02 .75 .11 .06 .74 .15
BS .03 .71 .11 .07 .66 .16

Note. T comprises the PTS and FCB-TI scales, S = SSS-V, L = Lie scale from the EPQ-R. For
description of scales see Table 4.3.

Table 4.5. Factor analysis based on EPQ-R, EAS-TS, DOTS-R and NEO-FFI scales

Model
Scale 1: PEN+T 2: PEN+T+L 3: PEN+T+F 4: PEN+T+L+F
I II III I II HI I II IH I II III

P .15 -.13 -.23 .17 -.11 -J7 .16 -.10 -.61 .12 -.13 -M
E .71 -.18 -.01 .72 -.17 -.09 .75 -.20 -.17 .74 -32 -2\
N -.09 .77 -.10 -.10 .77 -.11 -.06 .79 -.01 -.05 M -.04
L -.13 -.18 .46 -.14 -.11 .69
Soc .64 -.10 .02 .64 -.11 -.00 .66 -.11 .01 .65 -.12 .01
Act .69 .10 .19 .70 .11 .16 .69 .06 .24 .71 .06 .18
F -.17 .76 .02 -.19 .75 .06 -.15 .75 .13 -.14 .75 .13
D -.12 .82 .05 -.12 .82 .04 -.08 .80 .06 -.07 .80 .01
A .27 .66 .04 .28 .67 -.02 .30 .62 -.11 .30 .61 -.17
A-G .74 .15 -.00 .73 .14 .00 .71 .13 .06 .71 .12 .06
A-S .21 .26 -.23 .20 .26 -.25 .19 .23 -.20 .18 .22 -.21
A-W .62 -.18 .02 .61 -.19 .05 .56 -.22 .22 J7 -.21 .21
F-R .04 -.44 -.02 .04 -.44 -.01 .01 -.44 .16 .02 -.43 .14
MQ .68 .02 -.03 .67 .01 -.03 .64 -.02 .04 .64 -.02 .03
R-S -.13 .06 .78 -.09 .08 .76 -.07 .10 J5 -.03 .13 .45
R-E .10 -.04 .72 .15 -.01 .66 .14 -.00 38 .18 .01 .28
R-H .04 .14 .69 .09 .17 .61 .11 .19 35 .14 .20 .22
Dist .15 -31 .45 .17 -.30 .44 .13 -.28 39 .16 -.27 32
Pers .23 -.12 .38 .24 -.12 .41 .18 -.11 .53 32 -.08 .47
N -.27 .77 .11 -.26 .77 .10
E .80 -.17 -.03 .80 -.18 -.07
O .03 -.12 .02 .04 -.11 .01
A -.13 -.22 Jl -.11 -.19 J9
C .11 -.14 .69 .15 -.10 .73

Note. T comprises the EAS-TS and DOTS-R scales, F = NEO-FFI scales. The latter inventory
includes the following scales: N = Neuroticism, E = Extraversion, O = Openness, A =
Agreeableness, C = Conscientiousness. For description of the remaining scales see Table 4.3.
 Temperament and personality: Eysenck's three superfactors 85

The results depicted in Table 4.5 show that, independently of which kind of
scales are included in the factor analysis, two of the superfactors—extraversion
and neuroticism—can be identified. In all four models, the E and N scales have
loadings varying from 0.71 to 0.75 on the adequate factor identified as
extraversion, and from 0.77 to 0.80 for N on the neuroticism factor.
The conclusion regarding the stability of the E and N factors does not extend
to psychoticism. In Model 1, where PEN was combined with the EAS-TS and
DOTS-R scales, the P scale gave the highest loading on Factor III, but the level
of —0.23 can be regarded as marginal. The reason why psychoticism did not
come to light depends basically on the fact that P was suppressed by the three
rhythmicity scales from DOTS-R. When the Lie scale, which has much in
common with P, was included in the analysis, the loading of the P scale on
Factor III increased to —0.37. This result, however, does not allow us to
identify this factor as psychoticism. Only when the NEO-FFI scales were
included in the principal component analysis did psychoticism come out as a
marker (—0.61), and its loading increased when the L scale was included
(—0.66). The increase in loadings observed across Models 1-4 with respect to
the P scale may be regarded as a function of the number of psychoticism-
related scales included in the analysis.
The results based on sample C refer to the largest number of scales included
in the factor analysis. Again, all four models confirm that, independent of the
number of scales factor analyzed, there is a consistent stability with respect to
the occurrence of factors identified as E and N. The loadings on the adequate
scales vary from 0.71 to 0.75 for extraversion and from 0.77 to 0.80 for
neuroticism (see Table 4.6).
With respect to psychoticism, such stability in the factor structure, com-
prising different temperamental and personality scales as stated for E and N,
could not be confirmed. The factor analyses conducted for Models 1-4 confirm
that the loadings of the Psychoticism scale are a positive function of the
number of psychoticism-related scales included in the analysis. In Model 1 the
highest loading of the P scale (-0.40), present on Factor III, does not allow us
to identify this factor as psychoticism. It is mainly saturated by loadings of the
rhythmicity scales. The loading of P increased (-0.46) in Model 2 when the L
scale was added to the analysis. A further increase in the loading of the P scale
on Factor III occurred when the NEO-FFI scales were added (-0.54), and the
highest loading took place in Model 4 with all scales present (—0.60).
In summary, it can be stated that all 12 factor analyses (three sets of
inventories with four models each), in which different temperamental
inventories were included, and the NEO-FFI added, show beyond reasonable
doubt that, in a three-factor solution, two of the Eysenckian superfactors—
extraversion and neuroticism—come out independently of which tempera-
mental scales are taken into account. The E and N may be regarded as having
an universal status across scales measuring temperament in adults. The
86 Personality

Table 4.6. Factor analysis based on EPQ-R, PTS, FCB-TI, EAS-TS, DOTS-R and NEO-FFI scales

Model
Scale 1: PEN+T 2: PEN+T+L 3: PEN+T+F 4: PEN+T+L+F
I II III I II III I II III I II III

P .08 .18 -.40 .05 .22 -.46 -.11 .18 -.54 .11 .18 -M
E .16 .79 -.06 .14 .79 -.07 -.16 .81 -.08 .16 M -.10
N -.80 -.12 .06 -.79 -.13 .05 XI -.10 .04 -JB -.10 .03
L .36 -.27 .34 .31 -.25 .50
SE .68 .33 .17 .67 .35 .16 -.66 .32 .21 .65 .34 .18
SI .58 -.23 .18 .58 -.23 .20 -55 -.26 33 55 -.24 J4
MO .55 .38 .09 .54 .39 .08 -.54 .36 .14 53 .38 .11
BR .56 .37 .15 .55 .38 .14 -.54 .35 .21 53 .36 .17
PE -.72 .01 .06 -.71 -.00 .06 .72 .02 .10 -.73 .02 .09
SS .01 .22 .27 .01 .22 .26 .02 .20 .36 -.03 .22 32
ER -.82 -.27 -.03 -.81 -.28 -.02 .82 -.25 -.05 -.81 -.26 -.04
EN .73 .20 .10 .72 .21 .10 -.70 .18 .19 .69 .19 .18
AC .24 .75 -.06 .23 .75 -.06 -.25 .75 -.07 .25 .74 -.08
Soc .07 .56 -.07 .08 .54 -.04 -.11 .57 -.08 .11 .55 -.04
Act .13 .62 .10 .12 .61 .10 -.13 .63 .12 .13 .62 .11
F -.71 -.16 .07 -.70 -.17 .07 .72 -.16 .04 -.72 -.16 .04
D -.75 -.22 .02 -.75 -.23 .01 .77 -.21 -.03 -.77 -.21 -.05
A -.64 .22 -.01 -.65 .21 -.02 .64 .21 -.08 -.63 .21 -.11
A-G -.00 .61 .05 -.00 .61 .06 -.00 .60 .09 -.00 .60 .08
A-S -.24 .33 -.15 -.24 .32 -.15 .21 .29 -.11 -.21 .29 -.12
A-W .33 .62 -.00 .32 .62 .00 -.33 .60 .07 .32 .61 .05
F-R .51 .29 -.26 .50 .30 -.26 -.52 .29 -.18 .52 .29 -.18
MQ .13 .63 .07 .13 .61 .09 -.13 .61 .18 .12 .61 .18
R-S -.05 -.11 .66 -.06 -.10 .66 .08 -.12 .50 -.09 -.09 .49
R-E .02 .06 .62 -.00 .09 .59 -.01 .04 .43 -.01 .07 J8
R-H -.09 -.03 .56 -.10 -.02 .56 .11 -.05 .44 -.12 -.03 .42
Dist .20 .11 .55 .17 .14 .52 -.14 .08 .52 .12 .13 .45
Pers .19 .08 .58 .18 .09 .58 -.12 .08 .62 .11 .11 59
N .78 -.22 -.00 -.78 -.22 -.01
E -.23 .82 .01 .23 .81 .02
O -.05 .09 .22 .05 .11 .18
A -.16 -.15 .39 .16 -.16 .48
C -.20 .16 .61 .20 .17 .64

Note. T comprises the PTS, FCB-TI, EAS-TS and DOTS-R scales. F = NEO-FFI. For description
of scales see Tables 4.3 and 4.4.

Psychoticism scale, when related to other temperamental scales, does not occur
as a significant marker. Psychoticism comes to light only when accompanied, in
a factor analysis, by scales which have some elements in common with this
Eysenckian superfactor. The results suggest that the loading of the Psycho-
ticism scale in a factor solution is a positive function of the number of psy-
choticism-related scales present in that analysis. The data, based on principal
component analysis, suggest that P has a status different from E and N, and
they question the assumption that psychoticism has a unequivocal tempera-
mental nature.
 Temperament and personality: Eysenck's three superfactors 87

5. CONCLUSIONS

Several steps were taken to locate the three Eysenckian superfactors among
other temperament concepts. First, the psychometric characteristics of the
Polish version of the EPQ-R were presented. They confirm that the Polish
version of this inventory fulfills the requirements approved by Eysenck et al.
(1985). Second, the EPQ-R scales, taken as measures of the PEN factors, were
related to a variety of temperamental scales. Correlational data have indicated
that, among the 37 temperamental scales, 32 were correlated with E or N (or
with both), but not with P, if the level of >.30 is taken as a criterion. P was
related only to the SSS-V scales. Third, we conducted factor analyses of
temperamental scales, representing basic approaches to be met in research on
temperament, and took into account a variety of scale compositions. In these
analyses, the EPQ-R scales were always used as markers of PEN. The results
indicated that, whatever the configuration of temperamental scales being
included in the factor analysis, the E and N scales came always out with high
factor loadings. This suggests that extraversion and neuroticism (emotionality)
are stable factors across the instruments being applied for assessing tem-
perament. The factor loadings of the P scale differed, depending on what
configuration of temperamental traits were taken into account. P came to light
when few scales were included in the factor analysis (Model 1 in sample A) or
when the scales taken into consideration show some links to psychoticism, as
for example do sensation seeking, strength of inhibition (control of behavior),
conscientiousness, agreeableness, or the L scale.
Overall, the data indicate that two of the Eysenckian personality super-
factors—extraversion and neuroticism—are of a temperamental nature. They
appear always when they are related to other temperament constructs. In
contrast, psychoticism shows no or only weak links to temperamental traits.
The links occur mainly when the characteristics have some common elements
with, say, sensation seeking. The configuration of psychoticism, among the 37
temperament characteristics being measured in this study, suggests that
psychoticism is a personality factor different from these characteristics. When
considering personality, in accordance with Allport, as composed of two
subdivisions—temperament and character, it seems likely that psychoticism
belongs rather to the domain of character.

REFERENCES

Allport, G. W. (1937). Personality: A psychological interpretation. New York: Hold.

Anokhin, P. K. (1978). Collected works: Philosophical aspects of the functional system
theory. Moscow: Nauka (in Russian).
88 Personality

Barrett P., & Eysenck, S. B. G. (1984). The assessment of personality factors across 25
countries. Personality and Individual Differences, 5, 615-632.
Bates, J. E., & Wachs, T. D. (Eds.) (1994). Temperament: Individual differences at the
interface of biology and behavior. Washington, DC: American Psychological
Association.
Brebner, J., & Stough, C. (1993). The relationship between the structure of
temperament and extraversion and neuroticism. Personality and Individual Differ-
ences, 14, 623-626.
Buss, A. H., & Finn, S. E. (1987). Classification of personality traits. Journal of
Personality and Social Psychology, 52, 432-444.
Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits.
Hillsdale, NJ: Erlbaum.
Cattell, R. B. (1965). The scientific analysis of personality. Harmondsworth: Penguin
Books.
Claridge, G. (1987). Psychoticism and arousal. In J. Strelau, & H. J. Eysenck (Eds.),
Personality dimensions and arousal (pp. 133-150). New York: Plenum Press.
Claridge, R. Robinson, D. L., & Birchall, P. (1985). Psychophysiological evidence of
'psychoticism' in schizophrenics' relatives. Personality and Individual Differences, 6,1-
10.
Costa, P. T., Jr., & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-
R) and NEO Five Factor Inventory (NEO-FFI): Professional manual. Odessa, FL:
Psychological Assessment Recources.
Drwal, R. L. (1995). Adaptacja kwestionariuszy osobowosci [Adaptation of personality
questionnaires]. Warszawa: Wydawnictwo Naukowe PWN.
Eaves, L. J., Eysenck, H. J., & Martin, N. G. (1989). Genes, culture and personality: An
empirical approach. New York: Academic Press.
Eysenck, H. J. (1947). Dimensions of personality. London: Routledge & Kegan Paul.
Eysenck, H. J. (1952). The scientific study of personality. London: Routledge & Kegan
Paul.
Eysenck, H. J. (1956). The questionnaire measurement of neuroticism and extraversion.
Revista de Psicologica, 54, 113-140.
Eysenck, H. J. (1957). The dynamics of anxiety and hysteria. London: Routledge &
Kegan Paul.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Thomas.
Eysenck, H. J. (1970). The structure of human personality (3rd ed.). London: Methuen.
Eysenck, H. J. (1972). Human typology, higher nervous activity, and factor analysis. In
V. D. Nebylitsyn, & J. A. Gray (Eds.) Biological bases of individual behavior (pp. 165-
181). New York: Academic Press.
Eysenck, H. J. (1990). Biological dimensions of personality. In L. A. Pervin (Ed.),
Handbook of personality: Theory and research (pp. 244-276). New York/London:
Guilford Press.
Eysenck, H. J. (1992a). A hundred years of personality research, from Heymans to modem
times. Twelfth Duijker Lecture. Houten: Bonn Stafleu Van Loghum.
Eysenck, H. J. (1992b). The definition and measurement of psychoticism. Personality
and Individual Differences, 13, 757-785.
Eysenck, H. J. (1994). The Big Five or Giant Three: Criteria for a paradigm. In C. F.
Halverson, Jr., G. A. Kohnstamm, & R. P. Martin (Eds.), The developing structure of
 Temperament and personality: Eysenck's three super/actors 89

temperament and personality form infancy to adulthood (pp. 37-51). Hillsdale, NJ:
Erlbaum.
Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual differences: A
natural science approach. New York: Plenum Press.
Eysenck, H. J., & Eysenck, S. B. G. (1964). The Eysenck Personality Inventory. London:
University of London Press.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality
Questionnaire (Junior & Adult). London: Hodder & Stoughton.
Eysenck, S. B. G., Eysenck, H. J., & Barrett, P. (1985). A revised version of the
Psychoticism scale. Personality and Individual Differences, 6, 21-29.
Eysenck, S. B. G., & Haapasalo, J. (1989). Cross-cultural comparisons of personality:
Finland and England. Personality and Individual Differences, 10, 121-125.
Gale, A. (1983). Electroencephalographic studies of extraversion-introversion. A case
study in the psychophysiology of individual differences. Personality and Individual
Differences, 4, 371-380.
Goldberg, L. R., & Rosolack, T. K. (1994). The big five factor structure as an integrative
framework: An empirical comparison with Eysenck's P-E-N model. In C. F., Jr.,
Halverson, G. A. Kohnstamm, & R. P. Martin (Eds.), The developing structure of
temperament and personality from infancy to adulthood (pp. 7-35). Hillsdale, NJ:
Erlbaum.
Gray, J. A. (Ed.). (1964). Pavlov's typology: Recent theoretical and experimental
developments from the Laboratory ofB. M. Teplov. Oxford: Pergamon Press.
Gray, J. H. (1991). The neuropsychology of temperament. In: J. Strelan & A. Angleitner
(Eds.) Explorations in temperament: International perspectives on theory and
measurement (pp. 105-128). New York: Plenum Press.
Gross, O. (1902). Die cerebrale Sekundarfunktion [The cerebral secondary function].
Leipzig: Verlag von F. C. W. Vogel.
Guilford, J. P. (1959). Personality. New York: McGraw-Hill.
Hanin, Y., Eysenck, S. B. G., Eysenck, H. J., & Barrett, P. (1991). A cross-cultural study
of personality: Russia and England. Personality and Individual Differences, 12, 265-
271.
Heymans, G. (1908). Uber einige psychische Korrelationen [About some psychological
correlations]. Zeitschrift fur angewandte Psychologic, 1, 313-381.
John, O. P. (1990). The "Big Five" factor taxonomy: Dimensions of personality in the
natural language and in questionnaires. In L. A. Pervin (Ed.), Handbook of
personality: Theory and research (pp. 66-100). New York: Guilford Press.
Kagan, J. (1994). Galen's prophecy: Temperament in human nature. New York: Basic
Books.
Loehlin, J. C. (1992). Genes and environment in personality development. Newbury Park,
CA: Sage.
Nebylitsyn, V. D. (1972). Fundamental properties of the human nervous system. New
York: Plenum Press.
Oniszczenko, W. (1995). Diagnoza temperamentu w ujciu genetycznej teorii Bussa i
Plomina: Kwestionariusz Temperamentu EAS-TS [The diagnosis of temperament
according to Buss and Plomin's genetic theory: The adult EAS Temperament
Survey]. Studio Psychologiczne, 33, 97-111.
90 Personality

Pavlov, I. P. (1951-1952). Complete works (2nd ed.). Moscow & Leningrad: SSSR
Academy of Sciences (in Russian).
Pervin, L. A. (1993). Personality: Theory and Research (6th ed.). New York: Wiley.
Prior, M. R., Crook, G., Stripp, A. Power, M., & Joseph, M. (1986). The relationship
between temperament and personality: An exploratory study. Personality and
Individual Differences, 7, 875-881.
Ruch, W. (1994). Temperament, Eysenck's PEN system, and humor-related traits.
Humor, 7, 209-244.
Ruch, W., Angleitner, A., & Strelau, J. (1991). The Strelau Temperament Inventory-
Revised (STI-R): Validity studies. European Journal of Personality, 5, 287-308.
Rusalov, V. M. (1989a). Object-related and communicative aspects of human
temperament: A new questionnaire of the structure of temperament. Personality
and Individual Differences, 10, 817-827.
Rusalov, V. M. (1989b). The natural prerequisites of development of human
individuality. Studia Psychologica, 31, 53-67.
Stelmack, R. M. (1990). Biological bases of extraversion: Psychophysiological evidence.
Journal of Personality, 58, 293-311.
Strelau, J. (1983). Temperament, personality, activity. London: Academic Press.
Strelau, J. (1996). The regulative theory of temperament: Current status. Personality and
Individual Differences, 20, 131-142.
Strelau, J. (in press,a). The contribution of Pavlov's typology of CNS properties to
personality research. European Psychologist.
Strelau, J. (in press,b). Temperament: A psychological perspective. New York: Plenum
Press.
Strelau, J., & Angleitner, A. (Eds.) (1991). Explorations in temperament: International
perspectives on theory and measurement. New York: Plenum Press.
Strelau, J., & Angleitner, A. (1994). Cross-cultural studies on temperament: Theoretical
considerations and empirical studies based on the Pavlovian Temperament Survey.
Personality and Invidivual Differences, 16, 331-342.
Strelau, J., Angleitner, A., Bantelmann, J., & Ruch, W. (1990a). The Strelau
Temperament Inventory—Revised (STI-R): Theoretical considerations and scale
development. European Journal of Personality, 4, 209-235.
Strelau, J., Angleitner, A., & Ruch, W. (1990b). Strelau Temperament Inventory (STI):
General review and studies based on German samples. In J. N. Butcher, & C. D.
Spielberger (Eds.), Advances in personality assessment, (Vol. 8, pp. 187-241).
Hillsdale, NJ: Erlbaum.
Strelau, J., & Eysenck, H. J. (Eds.), (1987). Personality dimensions and arousal. New
York: Plenum Press.
Strelau, J., & Zawadzki, B. (1993). The Formal Characteristics of Behaviour—
Temperament Inventory (FCB-TI): Theoretical assumptions and scale construction.
European Journal of Personality, 7, 313-336.
Strelau, J., & Zawadzki, B. (1995). The Formal Characteristics of Behavior-
Temperament Inventory (FCB-TI): Validity studies. European Journal of Personality,
9, 207-229.
Strelau, J., & Zawadzki, B. (1996). Temperament dimensions as related to the Giant
Three and the Big Five factors: A psychometric approach. In A. Brushlinsky, & T.
 Temperament and personality: Eysenck's three super/actors 91

Ushakova (Eds.), V. D. Nebylitsyn: Life and scientific creativity (pp. 260-281).

Moscow: Ladomir.
Teplov, B. M. (1964). Problems in the study of general types of higher nervous activity
in man and animals. In J. A. Gray (Ed.), Pavlov's typology: Recent theoretical and
experimental developments from the Laboratory of B. M. Teplov (pp. 3-153). Oxford:
Pergamon Press.
Thomas, A., & Chess, S. (1977). Temperament and development. New York: Brunner/
Mazel.
Tomaszewski, T. (1978). Tatigkeit und Bewusstsein: Beitrdge zur Einfuhriing in die
polnische Tdtigkeitspsychologie [Action and consciousness: Contribution to the
introduction to Polish theory of action]. Weinheim/Basel: Beltz Verlag.
Windle, M. (1989). Temperament and personality: An exploratory interinventory study
of the DOTS-R, EASI-II, and EPI. Journal of Personality Assessment, 53, 487-501.
Windle, M., & Lerner, R. M. (1986). Reassessing the dimensions of temperamental
individuality across the life-span: The Revised Dimensions of Temperament Survey
(DOTS-R). Journal of Adolescent Research, 1, 213-230.
Zawadzki, B. (1995). Uniwersalnosc PEN—trzech superczynnikow osobowosci:
charakterystyka psychometryczna polskej adaptacji "Zrewidowanego Inwentarza
Osobowosci H. J. Eysencka (EPQ-R)" [Universality of PEN—Three supercactors of
personality: A psychometric characteristic of the Polish Adaptation of the "Eysenck
Personality Questionnaire—Revised." Studio Psychologiczne, 33, 147-188.
Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal. Hillsdale,
NJ: Erlbaum.
Zuckerman, M. (1991). Psychobiology of personality. New York: Cambridge University
Press.
Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking.
New York: Cambridge University Press.
Chapter 5
Psychology and medicine

D. K. B. Nias

1. EARLY BEGINNINGS

Eysenck's contribution to the field known as Medical or Health Psychology

began with his observation that throughout recorded history, claims have been
made that temperament is linked with physical illnesses ranging from minor
ailments to serious diseases such as cancer. In order to build on this foundation
of observational, descriptive, and anecdotal evidence, he designed an empirical
study with the oncologist, David Kissen (Kissen & Eysenck, 1962). Patients
attending Kissen's chest clinic for assessment and diagnosis were asked to
complete an early form of the EPQ. Of these patients about half were
subsequently found to be suffering from lung cancer. The main finding was
that this cancer group had significantly lower N scores than the noncancer
group. This fitted in with the early observational evidence that cancer patients
tend to be "unassertive, compliant people, inexpressive of negative emotions
like anger, fear and anxiety."
The above empirical finding was important because it was consistent with
the view that personality may play a role in the development of cancer and so,
in a true scientific spirit, Eysenck attempted further research. In his
autobiography, Eysenck (1990) describes how resistance and even hostility
among the medical establishment effectively put paid to any further research at
this time. It was not until his collaboration with Ronald Grossarth-Maticek
some 20 years later that a fresh start was made. In the meantime, while
continuing with his thinking about etiological factors in disease, Eysenck
became involved in other areas of medical psychology. These included diet and
its effect on intelligence and health, and how smoking interacts with other
etiological factors in cancer and coronary heart disease (CHD). The history of
his work on smoking has been presented in a book (Eysenck, 1991a) and has
been widely publicized and evaluated (e.g., Burch, 1986; Gilbert, 1995).
Instead of reviewing this work, the focus of this chapter is on the link between
personality and cancer with special reference to his collaboration with
 Psychology and medicine 93

Grossarth-Maticek. In addition, a brief evaluation of his work on dietary

supplementation is also presented, again with special reference to his yet to be
published collaboration with Grossarth-Maticek.

2. PSYCHOLOGICAL FACTORS IN CANCER

Although Eysenck was temporarily unable to pursue his work on cancer, other
researchers began reporting similar results to those obtained in Kissen's clinic.
Kissen himself, before his untimely death, conducted further studies to check
on the original result and to investigate how personality might interact with
other risk factors such as smoking. He was one of the first to argue for a
synergistic relationship between smoking and personality, claiming on the basis
of his results that "the poorer the outlet for emotional discharge the less the
exposure to cigarette smoke required to induce lung cancer" (Kissen, 1964).
One example of early research that is often cited is by Le Shan (1966). After
working for many years with cancer patients, he observed that "loss of hope"
was a characteristic coping style. About 70% of his cancer patients were
classified in this way compared with only 10% of other patients. While such
studies have been criticized for being observational rather than empirical,
studies using more rigorous methodologies have produced similar results. An
example is provided by Schmale and Iker (1971), who followed up 68 female
patients who had been tested for abnormal cervical cells and were subsequently
referred for a biopsy. Comparison in terms of psychological profiles was made
between 28 patients who were found to have cancer and 40 who were found not
to have cancer. Interviews prior to diagnosis were designed to detect
"hopelessness potential and/or a reaction of hopelessness six months prior to
the first abnormal (suspicious) smear." Based on this psychological factor,
diagnosis was accurately predicted in 50 of the 68 women—a 74% rate against
the 50% expected by chance.
Another study with a methodology similar to the above is by Cooper and
Faragher (1993) who studied 1596 women with symptoms of breast lumpiness
or tenderness. Prior to diagnosis they were assessed for personality and coping
style and compared with 567 symptom-free controls. Consistent with the
original result obtained with Kissen, it was found that with respect to
interpersonal difficulties "denying the existence of the problem proved to be
counterproductive, being associated with an increased risk of cancer."
One of the main criticisms of these research reports has been that they were
based on patients who were already ill or at least just about to be diagnosed.
Rather than personality or coping style being an etiological factor, it is of
course possible that the impact of diagnosis or even worrying about cancer
before it is discovered is sufficient to cause a change in personality or coping
style. It is therefore important to consider prospective studies in which people
94 Personality

are assessed for personality long before there is any suspicion of cancer. One
study that did this was started 50 years ago and involved 1300 medical students
being assessed for personality and other variables and then being followed up
ever since (e.g., Taylor, 1976). Although the study was intended to investigate
etiological factors in CHD, the most striking finding was that students
described as "loners" who suppressed their emotions "beneath a bland
exterior" were found to be 16 times more likely to have developed cancer than
those who gave vent to their emotions (Shaffer, Graves, Swanck, & Pearson,
1987). Although several measures of personality were taken, the main one was
the Rorschach—now known to have low reliability—and perhaps for this
reason the study has been largely overlooked.
Although prospective studies are useful for identifying possible causal
factors, a complementary design that is also useful is to assess the effect of
interventions on prognosis. A much publicized study to do this was by Spiegel
Bloom, Kraemer, and Gottheil (1989) at Stanford who worked with patients
with metastases from breast cancer. Having found that psychological inter-
ventions such as group therapy and training in self-hypnosis helped to improve
quality of life, they conducted an experiment with 86 patients to see if this had
any effect on survival times. Spiegel was "stunned" by the results, being cited as
saying "I nearly fell off my chair" and "I just couldn't believe it" (Barinaga,
1989). While the control group who received standard medical care lived for an
average of 19 months, the experimental group lived for an average of 37
months. Many suggestions have been made as to just how the psychological
interventions might have helped. For example, Oliver (1989) noted that the
control group received less radiotherapy. Perhaps those who received the extra
support were more likely to persevere with the radiotherapy, with this being
the important factor in survival time. This type of explanation may provide a
partial explanation but seems unlikely to account for the size of the difference,
which for many oncologists has remained a medical mystery.
Supporting evidence for psychological factors having an effect on prognosis
was already becoming apparent from a series of studies conducted across the
road from Eysenck's base at the Maudsley Hospital. In their main study at
King's College Hospital, Greer, Morris, Pettingale, and Haybittle (1990)
assessed coping style in 69 women after surgery for breast cancer. Those
classified as showing "fighting spirit" or "denial" were found to have fared
much better at follow up over intervals of up to 15 years than those showing
"stoic acceptance" or "hopelessness."
Another important study that warrants mention was conducted by Kneier
and Temoshok (1984). Patients with cancer were compared with those with
CHD and with a control group of healthy volunteers in their responses to
emotional phrases presented on slides. These slides included statements such
as "You deserve to suffer" and each time psychophysiological measures such
as skin conductance were taken and the subject asked if the statement
 Psychology and medicine 95

"bothered them." Subjects were scored as "repressive" types if they denied

being bothered but had autonomic arousal reactions. Cancer patients tended
to respond in this way while CHD patients tended to react in the opposite way,
with the healthy controls in the middle. Because of the different methodology
adopted in this study, it provides a useful source of evidence to complement
that from more conventional research designs. A comprehensive review of the
etiological and prognostic evidence, including Temoshok's own work but
excluding Grossarth-Maticek's, has been provided by Temoshok and Dreher
(1992). The above overview covers mere highlights, and although results from
other studies have mostly been less striking, the overall picture seems to be that
cancer-prone personality traits do exist and that appropriate interventions can
help to improve prognosis. The personality type may be summarized as
suppression of emotion and poor coping strategies leading to feelings of
hopelessness. Effective interventions seem to be those that lead to positive
changes such as an increase in social support. It is against this background that
it may be helpful to evaluate Eysenck's collaboration with Grossarth-Maticek's
extensive research program.

3. COLLABORATION WITH GROSSARTH-MATICEK

3.1 Prospective studies

Eysenck came across an on-going prospective study in Yugoslavia in which a
psychiatrist, Ronald Grossarth-Maticek, had in 1965 started assessing people
aged from 48 to 68 years for personality and other variables and then checking
on their state of health many years later. For those who had died he obtained
copies of the death certificate for recording causes of death. One of the best
predictors of death from cancer was based on a Rationality/Antiemotionality
interview schedule. It was being found that those who developed cancer tended
to have a characteristic personality as well as other more usual risk factors such
as smoking and stressful life events. The characteristic personality type was
described as follows: "they refused to admit to feelings of fear and anxiety, they
pretended that their life was governed entirely by rational motives, and that
intellect was more important than emotion" (Eysenck, 1990). This seemed very
much in line with the low N result obtained with Kissen.
Apart from supporting the original finding of a link between cancer and low
N, this study also provided evidence of a synergistic relationship between the
various risk factors. It was apparent that a combination of risk factors, such as
personality and smoking, was much more predictive of cancer than any one risk
factor alone. This agreement with Eysenck's own expectations and results
prompted him in 1980 to visit Grossarth-Maticek at his home which was now in
Heidelberg. They were able to compare notes on the problems they had each
encountered in their research, not least the resistance of the medical
96 Personality

establishment to pursuing their findings. Grossarth-Maticek had funded his

research by donations from relatives, having been unable to obtain grants from
official sources such as the Cancer Society. Going over the enormous amount
of data collected, they decided to pool resources and collaborate on its
interpretation and publication.
Some European experts in oncology were found to be dismissive of
Grossarth-Maticek's work, making all manner of criticisms. For example, it was
alleged that someone without a University position would not be able to obtain
permission to copy death certificates. However, Eysenck was told by the
Oberburgermeister of Heidelberg that special permission had been given in
this case. Eysenck found that although there were weaknesses in the research,
most of the more serious criticisms made could not be supported and he noted
that those who were most familiar with the research tended to hold it in higher
esteem than those who were not so familiar. He found that Grossarth-Maticek
was keen for anyone to look at his data, to interview those who had worked
with him, and in various other ways was completely open about his work. This
led Eysenck to conclude that the research deserved to be treated seriously and
indeed that it represented something "unique in the annals of epidemiology."
In an attempt at replication, a second study had been started in Heidelberg
involving a fairly random sample but with people in the age range 40-60 years
overrepresented. For a third study, these people were asked to nominate
friends and relatives who they knew to be highly stressed such as from suffering
unemployment or being separated from a loved one. In all three studies,
people were assigned at the beginning to personality types according to
interview and questionnaire data. The cancer-prone type was outlined as being
characterized by "lack of autonomy, lack of emotional expressiveness, the
repression of anxiety and anger, and reactions of hopelessness and helplessness
in the face of interpersonal stress." Other classifications included those
thought to be CHD-prone, described as having "strong feelings of anger,
aggression and hostility when faced with interpersonal difficulties and
problems," and a "normal healthy" type judged to be "autonomous, capable
of expressing emotions, and coping with stress more or less successfully."
Having allocated people to one of these types at the start, their health status
was then ascertained at various intervals years later. In each of the three
studies it was the cancer-prone type that was most likely to have succumbed to
this disease, with the normal, healthy type the least likely. Results from the
stressed group in the third study were clearest, with the death rate being some
40% higher, thus providing evidence that stress really can kill.
Because people in each of the studies had provided details of their smoking
history, the predictive value of this variable could be investigated. The
association with death from cancer was found to be many times less for
smoking than for personality. While it was found that smoking on its own was
less effective in predicting cancer (or CHD) than was personality, an
 Psychology and medicine 97

association with disease was clearly apparent when smoking was combined with
stress and a vulnerable personality. This again was consistent with the findings
of a synergistic relationship first reported by Kissen (1964) many years earlier.

3.2 Intervention studies

Aware that prospective studies do not necessarily provide evidence of a causal
link, Grossarth-Maticek designed intervention studies as a complementary
form of evidence. The aim was to see if modifying a person's stress coping style
could reduce the chances of cancer occurring. If intervention was to have such
an effect, then combined with the prospective evidence it would add
considerable weight to the likelihood of any link being causal.
From the cancer-prone group in the Heidelberg studies, 100 people were
recruited for a trial of behavior modification with half serving as a no-
treatment control group. In this way, 50 people received some 25 hours of
individual therapy aimed at making them more autonomous especially with
regard to coping with interpersonal problems and difficulties. At follow-up 13
years later, any deaths from cancer or other causes were recorded from death
certificates. Compared with 31 deaths from any cause in the control group,
only 5 had died in the therapy group. With respect to deaths specifically from
cancer, the figures were even more striking at 16 and zero, respectively.
In a second trial, 245 matched pairs of cancer (or CHD) prone people were
assigned to a therapy or control group. Therapy was along similar lines
designed to increase autonomy and improve stress coping but was conducted in
groups of about 25 at a time. Eight years later, 108 had died in the control
group compared with only 48 in the therapy group.
In a third trial, 600 people were given autonomy training and compared with
500 in a no-treatment control group and 100 in a placebo group. The placebo
treatment was based on psychodynamic principles which according to
Eysenck's (1952) outcome review would have no therapeutic value! In addi-
tion to group therapy sessions, each participant was given bibliotherapy in the
form of a leaflet outlining the techniques of autonomy training (bibliotherapy)
or in the case of the placebo treatment, "dynamic" advice. At follow-up, deaths
in the control group were 82% compared with 81% in the placebo group and
only 32% in the therapy group. Taken together, these three studies provide
strong evidence that appropriate interventions really can prevent disease
(similar results were obtained for CHD).
Grossarth-Maticek also designed studies to see if autonomy training could
prolong survival times in patients already suffering from cancer which had
spread. 24 matched pairs of cancer patients were divided into a therapy and a
control group. The control group lived for 3 years on average compared with 5
years for the therapy group. This finding it may be noted is similar to, although
not quite as striking, as that obtained at Stanford by Spiegel et al. (1989).
98 Personality

In another study, 50 women who had accepted chemotherapy were

compared with 50 who had rejected it. Half from each group were offered
autonomy training. The no treatment of either kind group survived for 11
months, those receiving chemotherapy only for 14 months, and those receiving
autonomy training only for 15 months. But the group receiving both chemo-
therapy and autonomy training survived for 22 months. This evidence of a
synergistic effect, along with the above study, was put forward as support for
the claim that an appropriate package of interventions can help to prolong life.
Taken together, these studies provide strong evidence for the efficacy of
psychological interventions in cancer. But this prompts an immediate question
of just what was it about Grossarth-Maticek's approach that could have had
such a dramatic effect? One possibility is that he somehow managed to get
people to actually follow his advice. In the context of low compliance rates
generally in medicine, having the ability to persuade patients to follow good
advice is an often overlooked skill for a therapist. Consistent with this
explanation, Eysenck (1996) observes in his latest review that Grossarth-
Maticek is "a charismatic type of personality." If improving compliance is a
major factor in therapeutic success rates, then it becomes an even more
important variable for researchers when designing outcome studies.

4. EVALUATION

Eysenck (1990) recognized that the above results would come as a surprise to
those who were unaware of previous studies, and who were used to viewing
body and mind as two separate entities. He reminds us that just as physicists
had to learn to think in terms of a space-time continuum, clinicians too will
have to learn to cope with a body-mind continuum. As already noted, the
results obtained from Grossarth-Maticek's extensive work are just part of a
large body of evidence supporting a two-way interaction between mind and
body. Several attempts have been made to evaluate his research and to
compare his results with those obtained by other researchers.
Recognizing that Grossarth-Maticek might be responsible for the most
important breakthrough in disease prevention since the discovery of penicillin,
a group of American researchers traveled to Heidelberg in 1988. They found
several inconsistencies in his papers, such as changes in the number of people
studied. This led one of them to conclude that "while these are not critical
mistakes, it does tell us that the guy is careless." Another, in noting some
statistical problems that cast doubt on the results, observed that "we have been
unable to achieve the level of certainty that science demands." An editorial in
Psychology Today (December, 1988) recounting this story, noted similar
shortcomings in great scientific discoveries in the past such as in Mendel's
 Psychology and medicine 99

genetic experiments. The conclusion from the American researchers was

inconclusive, with them ending by emphasizing the need for independent
replication as the acid test of scientific truth.
A special issue of Psychological Inquiry was devoted to an evaluation of the
research. In response to a target article by Eysenck, over 20 experts offered
their comments and analysis with a summing up reply to each being made by
Eysenck (1991b). This monograph provides a wonderful source of detail about
the research and is worth reading by anyone interested in the problems of
designing and conducting studies that will be taken seriously by critically aware
colleagues. A couple of examples may help to illustrate ways in which the work
was evaluated.
Amelang questioned whether there was clear separation between data
collection and subsequent analysis. To help reduce the chances of contami-
nation occurring once cause of death became known, Grossarth-Maticek took
the precaution of depositing lists of participants with their allocation to
personality types with two scientific institutes in Karlsruhe and Zurich.
Amelang claimed that this was unfortunately done 10 years after the study was
initiated, by which time cases of death were already occurring. Eysenck's
response was to point out that data were first deposited in 1977 with the
Oberburgermeister of Heidelberg, that cause of death was ascertained
exclusively through neutral investigators, and that Amelang's critique applied
anyway only to early data and so did not address the most recent findings.
A second example is provided by Vetter who has been a collaborator in the
analysis of the data for many years. He hypothesized that the team of
interviewers might have become aware of early signs of cancer (or CHD) when
interviewing people and that this might have influenced how they were
assigned to personality types. To test this hypothesis, Vetter compared results
obtained over several intervals of time. If such a bias was operating, then a
closer link between personality type and cancer would be expected when
diagnosis or death followed soon after assessment. Although there was some
evidence of this, the overall pattern of results led Vetter to reject his
hypothesis.
An extremely skeptical analysis of the research appeared in the British
Medical Journal. Pelosi and Appleby (1992) started by saying that the research
had been reported in "obscure or unrefereed journals and books," and then
directed their critique at reports in Behaviour Research and Therapy which
"contain some of the most remarkable claims ever to appear in a refereed
scientific journal." Turning to the questionnaire items, these were highlighted
as "clumsy" and lacking adequate data on reliability. Yet their predictive value
appeared to be "perhaps the highest ever identified in noninfectious disease
epidemiology." Pelosi and Appleby questioned whether participants could
have been inadvertently reassigned to personality types after cause of death
was known. Similarly, with regard to the intervention studies, they saw the
100 Personality

nature of the therapy involved as "fairly routine," which did little to explain its
outstanding success in preventing disease or increasing survival time. They
ended by calling for a reexamination of the original data by the Institute of
Psychiatry. Eysenck was not given the chance to respond at the time, but did so
in a subsequent issue.
Eysenck (1992) responded to the comment about obscure journals by listing
some of those involved which included the American Journal of Epidemiology.
In response to the criticism that the questionnaire items were too clumsy to
reveal any cancer- or CHD-prone types, Eysenck pointed out that the ques-
tions would have been clarified where necessary by the trained interviewers
who administered them. The accusation that participants were reassigned to
personality types once cause of death was known was pointed out to be both
unfounded and unworthy. Having to respond to such misconceptions led
Eysenck to wonder why they had not contacted Grossarth-Maticek or himself
before publishing their critique. Had they done so, some of their more
arbitrary accusations could have been cleared up so leaving them free to focus
on more substantive issues.
Eysenck described some of the ways in which the data collection had indeed
involved occasional problems. One example concerned the discovery that a few
questionnaires had been answered in an identical fashion by some subjects.
This may have been a coincidence or it could have been due to an interviewer
(over 100 students were involved) completing a few questionnaires on behalf of
the participants instead of conducting a proper interview. Because of such
doubts, the data were reanalyzed excluding any suspect questionnaires. It was
found not to matter because no systematic distortion became apparent.
Eysenck completed his response by questioning whether the results really were
"too good to be true." His answer was to cite results from other studies with
striking outcomes, such as the one already described by Spiegel et al. (1989).
In their response, Pelosi and Appleby (1993) returned to the issue of the
truly amazing nature of the results. They illustrated this by citing two claims.
First, that personality acted to "increase the risk of cancer by about 120 times
and the risk of ischaemic heart disease by about 25 times." Second, that
psychological interventions acted to bring about "massive reductions in death
rates over the next decade." Because of the importance of these claims, Pelosi
and Appleby repeated then" request for a critical reconsideration of the
research. In support of this request, they listed several aspects of the research
that need to be checked or elaborated. For example, given that "the remark-
able accuracy of prediction of cause of death years later would be impossible if
there were the slightest misclassification of personality types," they point out
that "more details are required on the training of the 100 or more workers who
collected the original and follow-up data, and on the reliability of the research
interviews in their hands." In the interests of designing replication studies, one
can but agree with the pressing urgency of their request.
 Psychology and medicine 101

5. FUTURE DIRECTIONS

In evaluating the evidence for a link between personality and cancer, it will be
of interest to analyze more carefully than has been done so far those studies
that have produced nonsignificant or even contrary results. Fox (1978) has
drawn attention to some of these exceptions and Scherg (1986) lists details of
40 separate studies. One obvious explanation for some of these inconclusive
results is that they were based on unreliable measures of personality, such as
the 550-item MMPI.
A recent example of an inconclusive result is provided by Amelang, Schimdt-
Rathjens, and Matthews (1996). In testing the theory of disease-prone person-
ality types, they collected data from 1858 people aged 39-68 years in the city of
Heidelberg. Rather than using Grossarth-Maticek's interview and question-
naire schedules, they designed their own battery of personality scales.
Unfortunately, their measures of cancer- and CHD-prone types were "inter-
correlated at the level of their reliabilities." Because of this it was hardly
surprising that a comparison of 58 people suffering from cancer with 208 cases
of CHD failed to detect any difference (although both groups scored differ-
ently from healthy participants). No doubt this study will be cited as a negative
result, even though it did not provide a fair test. This is a pity because the
authors plan to extend their study into a longitudinal one in an attempt to test
further the findings reported by Eysenck and Grossarth-Maticek. To their
credit they end their report by noting the fundamental design problem of
having "some differences in procedural details." As it stands, the study
provides an example of how not to conduct a replication study and as such will
provide further ammunition for those who point out the silliness of meta-
analysis (e.g., Eysenck, 1984).
Turning to positive results, the most dramatic evidence for the impact of
psychosocial factors on cancer comes from cases of recovery in patients
thought to be incurable. Weinstock (1977) in noting spontaneous regression of
cancer in hundreds of documented cases, looked to see what they had in
common. Where information was given about psychosocial factors, he noted
that "there was in each case a favorable change in the situation just preceding
the cancer shrinkage." Often the change took the form of a religious
conversion, falling in love, or suddenly finding a new purpose in life. Similarly,
Levoy (1989) refers to the work of O'Regan in California who has compiled a
database on spontaneous remission, covering some 3500 cases collected from
medical journals around the world. He suggests that in many of these cases,
psychosocial events somehow boosted the immune system to bring about a
cure. He also refers to the Lourdes International Medical Commission which
meets once a year in Paris to differentiate between miracles, hoaxes, and
spontaneous remissions. Note that this organization does not regard sponta-
neous remission as a miracle.
102 Personality

This type of evidence is often dismissed because it is anecdotal. Obvious

sources of error include mistaken diagnoses and patients responding parti-
cularly well to conventional treatments. Even so, it is tempting to think that at
least some of the thousands of recoveries are due to psychological factors.
Rather than continuing to ignore such anecdotal evidence, a rigorous study of
the phenomenon of spontaneous remission would be a promising next step. A
parallel can be made with the well-known placebo reaction. Rather than trying
to find out how much of a therapeutic intervention can be attributed to its
placebo value, it would be far better to investigate just what it is that produces
the placebo effect. In other words, we no longer need to be convinced about
placebo reactions but we do need to know how they operate.
Another promising approach would be to adopt Grossarth-Maticek's techni-
que of combining questionnaire measurement with an interview. In an analysis
of the accuracy of his predictions, it was found that the method of question-
naire administration was an important variable (Grossarth-Maticek, Eysenck,
& Barrett, 1993). The best predictions were achieved for subjects when both
trust and understanding had been increased by the application of suitable
interviewing skills. Trust was increased by talking to the participants for up to
an hour before administering the questionnaire. They were invited to talk
about life events and their typical reactions to these events. Then by dealing
with any doubts they might have, checks were made to help ensure that they
were happy with both the interviewer and the research. Understanding was
increased by reading aloud each question, and then responding to any queries
by trying to explain the precise meaning of the question. Based on the clearer
results obtained from using this approach, questionnaire researchers will now
need to decide how long to spend talking to their subjects!
Because the cancer-prone personality type appears to suppress emotions
rather than just being unemotional, a measure of emotionality on its own is not
enough. Rather than relying on scores such as low N for measuring suppres-
sion of emotion, an improvement is apparent from the work of Gudjonsson
(1981). Using a physiological measure of emotionality, he found that while
suppression of emotion was associated with low N, a much better predictor was
low N combined with a high L score. This more detailed type of assessment
seems likely to produce clearer and more consistent results in the future.
Yet another promising approach will be to direct future studies at younger
samples. Eysenck (personal communication) has noticed that studies involving
younger samples have more often tended to yield positive results when
compared with those involving older samples. This is not unexpected since
psychological factors are likely to have an earlier influence than the more
conventional physical risk factors which tend to be associated with aging.
In conclusion, a number of studies using different methodologies appear to
have arrived at a general delineation of a cancer-prone personality type. It also
appears from a limited number of intervention studies that training in stress
 Psychology and medicine 103

coping can help to both prolong the life of patients with cancer and to prevent the
development of the disease in the first place. Finally, there is both direct and
indirect evidence indicating the role of the immune system as the responsible
mechanism between psychological factors and cancer. Examples of evidence in
support of this promising area of investigation are provided by Levy, Horberman,
Maluish, Schlien, and Lippman (1985) and by Solomon (1987).
Because the studies reviewed in this chapter have generated a large amount
of criticism, sometimes warranted but not enough to explain away the results, it
will be some time before clinicians begin to take serious notice of the apparent
breakthrough. Even reviewers have sometimes avoided discussing the results
because they find them to be "simply unbelievable" (e.g., Fox, 1988). This is
unwise because even if errors and biases have been sufficient to distort the
findings, there is still much to be explained. Enough studies have been con-
ducted to form at least tentative conclusions, but as in other areas of science
this is not enough in the real world. It is only after many independent
replications and integration with existing theoretical principles that the basic
findings will have a chance of becoming established and incorporated in
clinical practice.

6. VITAMIN/MINERAL SUPPLEMENTATION

6.1 Introduction
One of the first studies to indicate the need for a dietary supplement was by
Harrell, Woodyard, and Gates (1955) who gave supplements to women of low
economic status in New York during pregnancy. Their children were tested for
IQ at 4 years of age and found to score 8 points higher than a control group of
children whose mothers had been given placebos during pregnancy. Benton
and Roberts (1988) and Schoenthaler (1991) administered supplements
directly to children and obtained similar results. In evaluating this work, it
was generally assumed that positive results had been obtained because some of
the individuals were seriously undernourished.
A BBC television documentary in 1988 presented evidence that more
children than previously thought were undernourished and that this was
adversely affecting their behavior and academic performance. In response to
these claims, the Dietary Research Foundation decided to fund further
research and invited Eysenck to join a panel of experts. A double-blind
experiment using vitamin/mineral and placebo capsules was organized by
Schoenthaler in both the U.K. and U.S.A. Reports on this work appeared as a
special issue in the house journal Personality and Individual Differences.
The main trial involved 400 children being divided into four groups and
receiving over a period of 12 weeks either a placebo or a supplement con-
taining 50%, 100%, or 200% of the U.S.A. recommended daily allowance. The
104 Personality

middle dose supplement produced the best results with scores on the WISC-R
being raised for nonverbal tests but not for verbal tests. Relative to the placebo
and practice effect baseline, the increase was of the order of 4 IQ points. In
trying to estimate the proportion of children who might have been under-
nourished, it was found that an improvement of 6 IQ points was obtained for
no less than 45% of the sample. The WISC-R was selected for the study
because it was predicted that any improvement in IQ would be limited to fluid
ability (nonverbal tests) rather than crystallized ability (verbal tests). In an
introduction to the reports, Eysenck (1991c) provided background information
and linked the research with earlier studies.
The next year, additional commentaries appeared as a special issue in The
Psychologist. In an introduction, Wardle (1992) refers to five placebo-
controlled studies which have reported beneficial effects from dietary
supplementation and two which have not obtained significant benefits. In
noting that no ready explanation has yet emerged for the different results, she
observes that positive results have been obtained when psychologists are
involved and nonsignificant results when the main researchers are nutritionists.
Perhaps this is an example of "experimenter effect" with psychologists being
more likely to believe that vitamins could affect IQ while nutritionists are likely
to be predisposed to doubt the idea.
Other explanations for the different conclusions from the seven placebo
controlled trials are apparent. In one of the inconclusive trials (Nelson,
Naismith, Burley, Gatenby, & Geddes, 1990), the supplements were admini-
stered for only four weeks. Compared with the 12-week period in the Dietary
Research Foundation study, this may have been too short a period to have
produced significant results. In the other inconclusive trial (Crombie et al.,
1990), the sample consisted of only 86 children. Even though a slight
improvement in IQ was obtained for the supplement group, the effect was not
quite significant. This borderline result has, unfortunately, been widely
reported as a negative result.

6.2 Grossarth-Maticek's Research

Grossarth-Maticek (unpublished) has collected data indicating the value of
taking a vitamin/mineral supplement when feeling the need, such as when
noticing the first signs of an infection. In a first study which started in 1982, he
recruited 810 people aged between 32 and 68 years who had suffered an illness
in the previous six months. They were randomly assigned to one of three
groups. Two of the groups were given a supply of vitamin/mineral capsules.
One of these groups was asked to take three to five capsules a day but only
when feeling the need. Those in the other group were urged to take either one
or two capsules per day on a regular basis. The third group served as a no-
treatment control. After three months, he checked to see how many people
 Psychology and medicine 105

had followed the advice, urged them to continue, and organized a long-term
follow-up study. After six years and again after 13 years, the subjects were
asked to report how often they had taken the capsules, how often they had
been ill, how often they had been off work, and so on.
People in the group who followed the advice to take a high dose when they
felt the need tended to be the healthiest. They were less likely to suffer
infections than the regular takers (results for one vs. two capsules a day were
very similar) or the control group. People who were the least healthy tended to
be those who were meant to be taking the capsules but who had given up.
Results for chronic illness after 13 years revealed that only 11% of the "high
dose as required" group reported being chronically ill compared with over 30%
for the regular takers, over 50% for the controls, and around 60% for those
who had given up. Other results such as for days off work followed a similar
pattern with the high dose as required group being healthier than the regular
takers, the controls, or those who had given up.
This study is important in that it involved a large sample taking a vitamin/
mineral supplement for many years. It did not include a placebo control, which
anyway would have been inappropriate over such a long time span. The results
are clear but because of the research design, alternative explanations do need
to be carefully considered. One alternative is that people whose health is
improving may be more likely to persevere with taking the capsules. Another is
that having something to do (taking the capsules) when noticing the first signs
of illness somehow gives a boost to the immune system. While such possibilities
cannot be ruled out, are they really sufficient to account for the whole
difference between the groups? This is clearly a case where the research must
be considered in the context of other studies such as placebo controlled ones.
In the second study, 328 patients with acute influenza were randomly
assigned to a treatment or a control group. Those in the treatment group were
given 50 capsules by their GP and asked to take three to five capsules a day for
10 days. The others were given standard advice and treatment. The main result
was that the vitamin group spent an average of only five days in bed compared
with 14 days for the control group.

7. FUTURE DIRECTIONS

An obvious next step will be to repeat the above study of influenza patients
with a placebo control. Such a study can then be added to the evidence from
the seven placebo controlled trials mentioned above. After that it will be
interesting to see what results are obtained for different age groups, people at
different fitness levels, different illnesses, and so on. Another major task will be
to discover which ingredients in the widely based supplement are responsible
for any therapeutic effects.
106 Personality

In conclusion, the evidence outlined above seems sufficient to challenge the

widely cited claim that a varied diet provides all the nourishment we need. This
notion has always seemed a strange one given the large number of nutrients
that the human body needs. Even if it were true, there would surely be a need
for supplementation when we are ill and not eating properly? Athletes appear
to be one group that has for many years played safe by adding supplements to
their diet, especially when in full training. If the above studies are successfully
replicated and if the results are widely publicized, then other groups of people
will be joining the athletes.

REFERENCES

Amelang, M., Schmidt-Rathjens, C, & Matthews, G. (1996). Personality, cancer and

coronary heart disease: Further evidence on a controversial issue. British Journal of
Health Psychology, 1, 191-205.
Barinaga, M. (1989). Can psychotherapy delay cancer deaths? Science, 246, 448-449.
Benton, D., & Roberts, G. (1988). Effect of vitamin and mineral supplementation in
intelligence of a sample of schoolchildren. Lancet, i, 140-144.
Burch, P. (1986). Smoking and health. In S. Modgil, & C. Modgil (Eds.), Hans Eysenck:
Concensus and controversy. Lewes, U.K.: Palmer Press.
Cooper, C. L., & Faragher, E. B. (1993). Psychosocial stress and breast cancer: The
inter-relationship between stress events, coping strategies and personality. Psycho-
logical Medicine, 23, 653-662.
Crombie, I. K., Todman, J., McNeil, G., Florey, C. du V., Menzies, I., & Kennedy, R. A.
(1990). Effect of vitamin and mineral supplementation on verbal and non-verbal
reasoning of children. Lancet, i, 744-747.
Eysenck, H. J. (1952). The effects of psychotherapy: An evaluation. Journal of
Consulting Psychology, 16, 319-324.
Eysenck, H. J. (1984). Meta analysis: An abuse of research integration. Journal of
Special Education, 18, 41-59.
Eysenck, H. J. (1990). Rebel with a cause. London: W.H. Allen.
Eysenck, H. J. (1991a). Smoking, personality, and stress. New York: Springer.
Eysenck, H. J. (1991b). Personality, stress, and disease: An interactionist perspective.
(Plus commentaries and response). Psychological Inquiry, 2, 221-232.
Eysenck, H. J. (1991c). Raising IQ through vitamin and mineral supplementation: An
introduction. (Plus reports). Personality and Individual Differences, 12, 329-333.
Eysenck, H. J. (1992). Psychosocial factors, cancer, and ischaemic heart disease. British
Medical Journal, 305, 457-459.
Eysenck, H. J. (1996). Personality and cancer. In C. L. Cooper (Ed.), Handbook of
stress, medicine, and health. Boca Raton, FL: CRC Press.
Fox, B. H. (1978). Premorbid psychological factors as related to incidence of cancer.
Journal of Behavioral Medicine, 1, 45-133.
Fox, B. H. (1988). Psychogenic factors in cancer, especially its incidence. In S. Maes, C.
D. Spielberger, P. B. Defares, & I. G. Sarason (Eds.), Topics in health psychology.
Chichester, U.K.: Wiley.
 Psychology and medicine 107

Gilbert, D. G. (1995). Smoking: Individual differences, psychopathology, and emotion.

Washington, DC: Taylor & Francis.
Greer, S., Morris T., Pettingale, K., & Haybittle, J. (1990). Psychological responses to
breast cancer: A fifteen-year outcome study. Lancet, i, 49-50.
Grossarth-Maticek, R., Eysenck, H. J., & Barrett, P. (1993). Prediction of cancer and
coronary heart disease as a function of method of questionnaire administration.
Psychological Reports, 73, 943-959.
Gudjonsson, G. H. (1981). Self-reported emotional disturbance and its relation to
electrodermal reactivity, defensiveness and trait anxiety. Personality and Individual
Differences, 2, 47-52.
Harrell, R. F., Woodyard, E., & Gates, A. I. (1955). The effects of mother's diet on the
intelligence of offspring. New York: Columbia University Press.
Kissen, D. M. (1964). Relationship between lung cancer, cigarette smoking inhalation
and personality. British Journal of Medical Psychology, 37, 203-216.
Kissen, D. M., & Eysenck, H. J. (1962). Personality in male lung cancer patients. Journal
of Psychosomatic Research, 6, 123-137.
Kneier, A. W., & Temoshok, L. (1984). Repressive coping reactions in patients with
malignant melanoma as compared to cardiovascular disease patients. Journal of
Psychosomatic Research, 28, 145-155.
Le Shan, L. (1966). An emotional life-history pattern associated with neoplastic disease.
Annals of the New York Academy of Science, 125, 780-793.
Levoy, G. (1989). Inexplicable recoveries from incurable diseases. Longevity, October.
Levy, S., Horberman, R. B., Maluish, A. M., Schlien, B., & Lippman, M. (1985).
Prognostic assessment in primary breast cancer by behavioral and immunological
parameter. Health Psychology, 4, 99-113.
Nelson, M., Naismith, D. J., Burley, V. Gatenby, S., & Geddes, N. (1990). Nutrient
intakes: Vitamin-mineral supplementation in British schoolchildren. British Journal
of Nutrition, 64, 13-22.
Oliver, R. T. D. (1989). Psychological support for cancer patients. Letter to Lancet, ii,
1209.
Pelosi, A. J., & Appleby, L. (1992). Psychological influences on cancer and ischaemic
heart disease. British Medical Journal, 304, 1295-1298.
Pelosi, A. J. & Appleby, L. (1993). Personality and fatal diseases. British Medical
Journal, 306, 1666-1667.
Scherg, H. (1986). Zur Kausalitatsfrage in der psychozialen Krebsforschung.
Psychotherapy and Medical Psychology, 36, 98-109.
Schmale, A. H., & Iker, H. (1971). Hopelessness as a predictor of cervical cancer. Social
Science and Medicine, 5, 99-109.
Schoenthaler, S. (1991). Improve your child's IQ and behaviour. London : BBC Books.
Shaffer, J., Graves, P., Swanck, R., & Pearson, T. (1987). Clustering of personality traits
in youth and the subsequent development of cancer among physicians. Journal of
Behavioral Medicine, 10, 441-447.
Solomon, A. (1987). Psychoimmunology: Interactions between central nervous systems
and immune systems. Journal of Neuroscience Research, 18, 1-9.
Spiegel, D., Bloom, J. R., Kraemer, H. C, & Gottheil, E., (1989). Effect of psychosocial
treatment on survival of patients with metastatic breast cancer. Lancet, ii, 888-891.
108 Personality

Taylor, C. B. (1976). Precursors of premature disease and death: The predictive value of
habits and family values. Annals of Internal Medicine, 85, 653-658.
Temoshok, L., & Dreher, H. (1992). Type C behavior and cancer. New York: Random
House.
Wardle, J. (1992). Nutrition and IQ (Plus reports). The Psychologist, 5, 399.
Weinstock, C. (1977). Notes on "spontaneous" regression of cancer. Journal of the
American Society of Psychosomatic Dentistry and Medicine, 24, 106-110.
Chapter 6
Psychoticism as a dimension of personality

S. Eysenck

I. INTRODUCTION

Extraversion (E) and Neuroticism (N) have been accepted as personality

dimensions for many years. Their measurement in questionnaire form was
perfected in the Maudsley Medical Questionnaire (Eysenck, 1956), then the
Maudsley Personality Inventory (Eysenck, 1959) and finally the Eysenck
Personality Inventory (H. J. Eysenck & S. B. G. Eysenck, 1964).
However, H. J. Eysenck suggested a third dimension of psychoticism (P) in
1952 in the Scientific Study of Personality, followed by some experimental
material outlined in Perceptual Processes and Mental Illness (Eysenck, Granger,
& Brengelmann, 1957). The next step clearly was to devise a scale to measure P
which proved to be a monumental task in view of the inherent problems in
expecting high P scorers to introspect—a near alien task for people with such a
personality profile!
The attempt to integrate P into the personality structure in questionnaire
form began with the PEN evolved into the PQ and several forms later was
published as the Eysenck Personality Questionnaire (EPQ, H. J. Eysenck & S.
B. G. Eysenck, 1975) with its final form being the Eysenck Personality Scales
(EPS, H. J. Eysenck & S. B. G. Eysenck, 1991).
Once a reliable and valid P scale was achieved, it became possible to
measure its concomitances and build up a picture of the characteristics of the
typical high P scorer. The measurement of the P dimension, of course, was
refined by factor analysis, as were N and E, but what now needed to be done
was to correlate this measure with objective tests. Many of these have been
listed in Psychoticism as a Dimension of Personality (H. J. Eysenck & S. B. G.
Eysenck, 1976) and the Manual of the EPS (H. J. Eysenck & S. B. G. Eysenck,
1991).
The latter manual lists studies concerning the P scale covering its factorial
stability, aggression, biological basis, delusions and hallucinations, inhibition,
reaction time, arousal, conditioning, visual perception, stimulus reduction and
response uncertainty, sex and sexuality, punitiveness, drug dependence and
110 Personality

health, interests and occupations, and miscellaneous other studies. It would,

therefore, be pointless to relist these earlier studies; suffice it to say that there
have now been appreciably many experimental studies to map out where the P
dimension is located as far as objective and biological tests are concerned.
Ever since we published the P scale, it has been widely used, debated,
criticized, and defended. It may be useful to indicate what precisely is meant by
saying that P is a dimension of personality. Most models of personality rest
content with factor analyses of trait constellations; the Big Five is a good
example of this tendency. Eysenck (1991) has argued that a taxonomy so
narrowly based is inevitably subjective, and does not contain any objective
criterion to help us decide between different models, such as those of Cattell,
Gray, Cloninger, and so on.
Figure 6.1 shows in outline the proper composition of a testable model, that
is, one which contains a theoretical basis from which deductions can be
derived, and which can be judged according to the validity of these deductions.
Such a model may begin with the postulation of certain personality dimensions,
go on to the factorial validation of the dimensions, as the central feature of the
model, but must introduce causal theories to anchor the dimensions more
firmly in a nomological network of distal and proximal antecedence, thus
generating predictions testable in the laboratory (proximal consequences) and

Genetic Biological Psychometric Experimental Social

Personalty kitoirooclortes Trait Studes Behaviour
Determinants Constelatlons
Figure 6.1. Diagramatic picture of the major constituents of a proper personality theory.
 Psychoticism as a dimension of personality 111

in the context of social behavior (distal consequences). Such a model must

begin with genetic determinants (DNA) and contain an account of biological
intermediaries that link DNA and the behavior that gave rise to the concepts of
personality dimensions. These biological intermediaries can then be used to
give rise to the prediction of proximal and distal consequences.
Figure 6.2 gives a diagrammatic view of the P concept. What is posited is a
continuum going from an altruistic, socialized, empathic, conventional, and
conformist sort of person to the other extreme of an impulsive, hostile,
aggressive, psychopathic, schizoid, and ultimately near-psychotic person. The
curve labeled PA indicates the probability that a person at any point of this
continuum will develop a functional psychosis if put under great stress; high P
itself merely constitutes an underlying personality profile that makes a person
more likely to develop psychotic symptoms.
The notion of such a continuum goes counter to the usual psychiatric
categorical system of taxonomy which asserts the existence of numerous mental
illnesses, qualitatively different from each other. Criterion Analysis (Eysenck,
1950, 1952) was worked out to decide on an experimental basis between the
dimensional and the categorical views, and supported the dimensional
approach quite decisively.
More recently, criterion analysis in the guise of the proportionality criterion
has been used to test the applicability of the continuum hypothesis further.

PA

AVERAGE

u
'£ E §
*
£ I
§ 1
.§
< </5
Figure 6.2. The continuum defining psychoticism. PA is the probability of a person at any point on
the continuum developing psychotic symptoms under stress.
112 Personality

The strong dependence of P on genetic factors has been documented in

Genes, Culture and Personality (Eaves, Eysenck, & Martin, 1989), and its
biological roots have been discussed at length by Zuckerman (1991). Of
particular interest here is the link between dopamine functioning, which is
closely linked with modern theories of schizophrenia and P (N. Gray,
Pickering, & J. Gray, 1994; see also chapters 1 and 3). This relationship has
given rise to renewed testable predictions, the most interesting one being a link
with creativity and genius.
H. J. Eysenck (1995, chapters 6 and 7) in his book on Genius, the Natural
History of Creativity has long chapters on the links between creativity and
psychoticism (and psychosis). Suffice it to add that one of the interesting
questions arising is whether there are potentially creative children who are not
high on psychoticism and therefore do not succeed in later life. Maybe the link
between creativity and psychoticism occurs because those without the
motivation and sheer negativistic persistence of high P scorers do not
succeed, however creative they are. In other words, could it be that the
disapproval of parents and teachers of creativity (being different and having
unorthodox ideas) in children who are low on P has the effect of stifling this
tendency? Conversely, perhaps the disapproval of parents and teachers may
spur high P scorers on and reinforce the pattern, hence allowing creativity to
flourish.
So much has by now been written about psychoticism, especially by H. J.
Eysenck, that the scientific and experimental facts are fully documented in
many books and articles. Here, I should like to take a more applied approach
and list some of the areas in which high P scorers display characteristic
behavior.

2. CHILDREN

H. J. Eysenck & S. B. G. Eysenck (1975) constructed a children's personality

questionnaire in which psychoticism was measured. Those children who scored
high on P were subsequently found to be those who had learning difficulties,
truanted, had behavior problems, got into crime, were loners, and were
disliked by staff and peers. Extraverted children by contrast were regarded as
"lovable rogues" and had their misdemeanors forgiven more readily.
Hence, although high P and high E children have a tendency to misbehave,
the difference appears to be that high E children do so out of high spirits and
create mischief, while high P children have a spiteful, disruptive type of
behavior, rather less beloved by teachers. Clearly the therapy should differ, but
alas, any learning or behavioral problems seem to be dealt with by the same
treatment, that is, some sort of child guidance or psychotherapy.
 Psychoticism as a dimension of personality 113

One of the practical advantages of being able to measure psychoticism is that

it is now possible to treat high P scoring children with a form of token economy
which has been developed most successfully by D. Lane, author of the book
The Impossible Child and whose M. Phil, thesis in 1977 dealt with Persistent
Failure and Potential Success: a Study of Failure to Respond to Therapy in
Children.
Basically, rather than dealing with noncooperative children with more and
more one-to-one therapy (which is anathema to high P scorers), D. Lane
entered into contracts with them, believing high P scorers live by the motto
"what's in it for me?" If suitably rewarded for returning to classes and
studying, the hope is that the reinforcement may cause these previously
isolated children to return to the fold.
Finally, although the same pattern of high P characteristics apply to child
offenders as can be seen in adult criminals, there does appear to be a different
personality pattern for adults and children. In 1975, using the EPQ, we item
analyzed both the adult and the junior versions to arrive at a criminality scale
(C). For adults, prisoners were asked to complete the EPQ and those items on
which the normal group differed from the replies of the prisoners with a high
degree of significance were listed to form the criminality scale. For children,
the same was done using two criteria. The first was a self-report antisocial
behavior questionnaire (ASB) devised by Allsopp and Feldman (1974, 1976)
which consists of 48 items of misbehavior ranging from very mild acts (e.g.,
making a noise in class) to serious offenses (e.g., breaking into private property
to steal something). The other criterion consisted of the number of detentions
given to the boys who formed the sample for this particular study during the
complete school term at the end of which the testing took place. The C score
was formed by using items on the junior EPQ which gave the best
discrimination between high and low scorers on the ASB.
The point of interest here is that the junior C scale has a preponderance of P
and E items while the adult C scale has mainly P and N items. Somehow,
youthful "naughtiness" is outgrown by high E scoring children while a high PN
pattern of the adult offenders suggests that neuroticism is the dimension that
appears to make juvenile offending a habit which persists. Hence, while P is
involved in misbehavior and criminality it clearly is of utmost importance to
look at the other personality factors possessed by a child if any therapy or
prediction is to be attempted.

3. THERAPY

Just as with children, adult high P scorers do not have a good prognosis with
conventional psychotherapy or counseling. There is a paranoid, suspicious
barrier as it were, which makes a one-to-one therapeutic relationship
114 Personality

impossible. Hence, it is high P patients who are documented as failing

appointments, resisting treatment, switching therapists frequently, and
generally taking longer time and more sessions before recovering (Rahman
& Eysenck, 1978).
High P scorers with a high N component are usually termed Personality
Disorders of one kind or another. One can visualize the effect of treatment as a
see-saw. When N goes down due to counseling or tranquilizers, the P
component rises. Released of his or her anxieties the self-assured aggressive P
side gains the upper hand and the patient drops out of therapy following his
usual loner, hostile lifestyle. This, of course, leads to conflict with family and
workmates thus arousing anxiety again, so that the patient seeks therapy
again—preferably with a different therapist! This see-saw effect means that
therapy is never really at an end and probably the answer is to deal
pharmacologically with the P component while trying to sort out the N type
problems. Unfortunately, even a mild dose of the major tranquilizers makes
patients feel "bad," i.e. when lowering P, anxiety gains precedence, so patients
are loathe to cooperate. In an ideal world, therapists would have the time and
patience to deal with patients who are anxious, having temporarily removed the
P "barrier," leaving the path clear for a conventional psychotherapeutic
approach.
While psychiatrists can tackle psychoses with psychopharmacological
interventions and can deal reasonably well with neuroses with Behavior
Therapy or conventional Psychotherapy, they are often baffled by the mixture
of P and N so characteristic of the personality disordered criminal, so much so
that the Mental Health Acts refer to Personality Disorders and Psychopaths as
"untreatable." Perhaps in due course, knowledge of the characteristics of
psychoticism can help to devise a viable treatment for these mental
illnesses.

4. IMPULSIVITY

Although the EPI had several items purporting to measure impulsiveness,

these always loaded on E but never to the same extent as did sociability (see
also chapters 4 and 10 in this volume). While sociability items loaded around
0.4 and above, impulsiveness loadings remained modestly around 0.2 or 0.3.
Until P was included, there was no appropriate factor for the impulsiveness
items to load on, but because there were certain items which consistently
loaded modestly on E and because these were certainly impulsiveness items
contentwise, they were retained on E in the EPI.
However, it was found that when the P dimension was included and the EPQ
was standardized, some impulsiveness items switched to P from E. We then
decided that impulsiveness was too important to be relegated to a small part of
 Psychoticism as a dimension of personality 115

the EPQ, removed the impulsiveness items from E, and began to refine an
impulsiveness questionnaire on its own. What then emerged was that
factorially there were two distinct aspects of impulsiveness, one aligning with
P and the other with E (S. B. G. Eysenck & H. J. Eysenck, 1978; Eysenck,
Pearson, Easting, & Allsopp, 1985). These two factors were named Impulsive-
ness (the P component) and Venturesomeness (the E component). Empathy
items (Mehrabian & Epstein, 1972) were included, initially as buffer items but
subsequently they turned out to be of considerable interest as a scale.
The junior version of the Impulsiveness Questionnaire (Eysenck, Easting, &
Pearson, 1984) confirmed the same set of intercorrelations as those of the
adult questionnaire, namely that impulsiveness correlated mainly with P and N
and modestly with E, while venturesomeness correlated mainly with E and
modestly with P. These intercorrelations are reproduced in Eysenck (1981)
where antisocial behavior was included and the subjects were children. Of
interest was the usual P alignment with impulsiveness and incidentally a
negative correlation of P with empathy, there being a positive correlation of N
with empathy as one would have expected. Moreover, P correlated very
decisively with antisocial behavior!
An analogy to describe the difference between the E type of impulsiveness
and the P type can be made by a driving example. Imagine a driver who
overtakes on a blind bend. The high P driver is impulsive and it never occurs to
him or her that someone might be coming the other way, nor that he or she
might then be involved in a fatal crash. The high E driver, on the other hand,
will be fully aware of the danger and will take the risk, simply for the sensation
seeking arousal the episode may cause. In short, impulsiveness of the PN
variety seems to be the pathological branch while the E sort could be regarded
more as a risk taking stance.

5. ADDICTION

Many books and articles have been written about addiction and much research
has concentrated on a social phenomenon which has become increasingly
difficult to deal with (Eysenck, in press). Given the links addiction has with
crime, some solution is urgently required and governments all over the world
have poured money and resources into various studies and projects, nearly all
to little avail. The problem is largely twofold—first, addiction fulfills a need for
high PN low EL type personalities, and second, drug dealing is a highly
lucrative and organized occupation.
Most studies, employing an individual differences approach, have concluded
that addicts of all kinds (hard drugs, alcohol, tobacco, bulimia, etc.) score high
on P (Gossop & Eysenck, 1980, 1983; Feldman & Eysenck, 1986; Teasdale,
116 Personality

Seagraves, & Zacune, 1971). The role of N is more problematic, since it may
often be a concomitant or consequence of being an addict, but the point is that
it is not a necessary part of addiction as P apparently is.
Take alcoholics, for example. Some alcoholics score high on P and N, but
others score high on E only. Why are they drinking to excess is the question
and studies using the EPQ have come up with interesting answers (Ogden,
Dundas, & Bhat, 1988). One could speculate that high N subjects drink to
drown their sorrows and anxieties, high P subjects drink to excess because they
fail to monitor when they have had enough, and high E subjects get carried
away by the social aspect of drinking, being unable to resist the "buying by
rounds" that is so popular in pubs.
Now, some years ago "control drinking" was suggested and tried in an
attempt to teach alcoholics how to control the level of their drinking. Alas, no
individual differences in drinkers were considered so that among the subjects
were high P scorers who are unable and unwilling to control their drinking
habits and who went on notorious binges. Immediately, control drinking
became a dirty and taboo phrase and the baby was promptly thrown out with
the bathwater! The rational answer must surely be that therapy should be
tailored to the personality of the alcoholic and control drinking should only be
used for high E subjects and in any case, never for those high on P.
Total abstinence is crippling socially and so only in extreme cases (high P
scorers) should this be the therapy of choice. It must always be remembered
that all addicts, whether using hard drugs, alcohol, or tobacco, do so because it
is rewarding in the short term and is a solution of choice for all sorts of crises.
If addicts come into conflict with the law, or lose their jobs or families as a
consequence of their habit, they may be forced to seek or accept treatment, but
deep down, this is their way of coping with crises and although a cure may be
possible, relapse is very much on the cards. Consequently, if it is possible to
avoid total abstinence and if a self-help group can monitor a new lifestyle, the
chances of relapse may well be reduced.

6. SELECTION

Personality tests are becoming more and more popular for all kinds of
occupations, but what is the ideal personality profile employers are all seeking?
High intelligence is undoubtedly desirable for most professions but not always
for routine jobs. Sociable, nonpathological characters are certainly most
pleasant to work with, but occupations differ in their required personality
traits. Salesmen need to be high E scorers, as are confidence tricksters,
whereas air controllers need the vigilance of introverts.
 Psychoticism as a dimension of personality 117

From a selection point of view, given the pathological connotations high P

scoring conjures up, presumably the prevention of high P scoring employees is
partly what selection tests are designed for. While it is desirable to give
selection tests in order to ascertain the personality of applicants, scoring high
on P or N is not always a disadvantage. Social workers, probation officers,
psychiatrists, and counselors do need a degree of empathy to understand their
clients and there is a persistent correlation between empathy and N. Those
employed in the "caring professions" do need some empathy or they would
seem lacking sadly in "bedside manner."
Similarly, there are occupations in which P is an advantage. For example,
when we asked student nurses to complete the EPQ, the manual (H. J.
Eysenck & S. B. G. Eysenck, 1975) shows that they had fairly high P scores.
The subsequent selection of fully trained nurses revealed that their P score
mean was higher, so presumably a certain amount of the P qualities are needed
in order that nurses can cope with some of the heart-rending cases they have to
deal with. Surgeons too need some of the cool, impersonal, unempathic
characteristics of high P scorers. We have mentioned creativity (Eysenck, 1995)
and its relationship with P, and here is an ideal area for high P scorers to excel
in or at least to gravitate to.

7. (FURTHER) ADVANTAGES OF PSYCHOTICISM

The notion of P being a "bad" thing in every way is not an accurate picture.
Hare (1980) remarked that psychopaths are physiologically in a most
advantageous position, since they are so equipped as not to feel guilt or
empathy and can, therefore, gain maximal pleasure and behave exactly as they
wish without remorse or any feeling of regret. Where they come unstuck, of
course, is in their co-existence with others in society who object to such
behavior, most civilizations being built on adherence to certain rules, laws, and
customs which allow people to live in reasonable peace with each other.
Moreover, when we attempted to locate high P scoring neurotics in an Isle of
Wight mental hospital, we were astonished to find that they were nearly all out-
instead of inpatients. One can only speculate that the reason for this is that
"pure" neuroses, particularly anxiety or depression, make people feel helpless
and hopeless enough to need inpatient treatment. Add a dash of P, however,
and it seems patients have a "coping mechanism" of some sort which allows
them to remain unhospitalized and needing outpatient treatment only. In other
words, P does in some respects appear to act like an antidote to N, in that
aggression, anger, and suspicion may be better for one's soul than anxiety, fear,
and hopelessness!
118 Personality

This then may well have been the argument the "antipsychiatrists" gave in
the 1960s for allowing schizophrenics to "do their own thing." In other words
to let them behave as they wanted and not to medicate them, thus rendering
them like "vegetables." This was all very well for schizophrenics but not too
good for all others, some of whom were in conflict with them, especially their
families.
Probably the conclusion to reach is that once it is established that someone
scores high on P, rather than or in addition to N, all kinds of consequences
follow. It is wise to choose a correct occupation, unwise to suggest control
drinking, contraindicated to use mere counseling or psychotherapy, token
economy proving to be a better bet.
Psychoticism should be regarded as a personality dimension, which without
N may well be confined to an eccentric lifestyle in high P scorers. When the
extra gene introduces schizophrenia, those patients who either seek or are
referred for psychiatric help invariably also have a sizable N component.
So whatever the advantages or disadvantages of possessing a high P score
may be, what is certain is that it is vital to measure this dimension (and others!)
and to know that we are dealing with a strongly inherited, not to be overlooked
personality characteristic which can be modified if recognized or encouraged in
appropriate circumstances.

8. DOES "PSYCHOTICISM" RELATE TO PSYCHOSIS?

It has often been objected that the P scale seems to be related to psychopathy
rather than to functional psychosis such as schizophrenia. It was to answer this
question experimentally that Eysenck suggested his method of proportionality,
which would also serve to answer the question of whether psychosis was really
the extreme end of a continuum, or was rather a separate taxon categorically
distinguished from normality (Eysenck, 1992). The method essentially uses
measures (psychological, physiological, hormonal, or neurotransmitter) that
clearly differentiate between normals and schizophrenics. Then it argues that
if, and only if, we are dealing with a continuum from normal to schizophrenic
would it also be true that high P normals and low P normals would be
differentiated by those measures with high P scorers scoring like schizo-
phrenics and low P scorers like normals. This is an adaptation of the method of
criterion analysis introduced earlier for the same purpose, but requiring a
much more complex experimental set-up (Eysenck, 1950).
In the list of variables used to illustrate the proportionality criterion, several
different types of measures have been included. One class deals with biological
variables (HLA B27, MAO; serotonin) of different kinds. A second deals with
laboratory behaviors (eye-tracking, dichotic shadowing, sensitivity levels). A
third is concerned with learning-conditioning variables (latent inhibition,
 Psychoticism as a dimension of personality 119

negative priming). Yet a fourth is concerned with psychological variables

(creativity, hallucinatory activity, word association). Physiological variables
(EMG, autonomic-perceptual inversion) constitute yet a fifth set of variables.
It is the variety of variables which makes the results impressive, together with
the theoretical congruence; to obtain successful results over such a wide array
of variables suggests that the underlying hypothesis may be along the right
lines.
Particularly important is the fact that variables closely associated with
theories of schizophrenia, such as dopamine D2, are also associated with P
(Gray et al., 1994). Such close links point quite clearly to an association
between P and schizophrenia; there is no similar evidence linking it with
psychopathy, unless it is agreed that psychopathy involves psychoticism and
may just be related to psychosis by having a lower position on the P continuum,
as Eysenck believes. This and many other important issues are clarified by the
theory underlying P and linking it with psychopathology, such as its close link
with genius (Eysenck, 1995) and addiction (Eysenck, in press). It would need a
good deal of contrary experimentation to counter the confirmed deductions
from the theory of psychoticism.

9. INTERPRETING PSYCHOTICISM IN ISOLATION FROM OTHER DIMENSIONS

The typical, extreme high P scorer is described in the EPQ-R manual (H. J.
Eysenck & S. B. G. Eysenck, 1991). He or she are listed as being solitary, not
caring for people, troublesome, not fitting in anywhere, cruel and inhumane,
lacking in feeling and empathy, insensitive, hostile even to his or her own kith
and kin, and aggressive even to loved ones. They have a liking for odd and
unusual things and a disregard for danger, enjoying making fools of others and
upsetting them. Socialization is a concept which is relatively alien to high P
scorers; empathy, feelings of guilt, sensitivity to other people are notions which
are strange and unfamiliar to them.
However, many studies dealing with individual differences look only at one
dimension at a time and hence often fail to find significant results. Take P for
example. If differences between high and low P scorers is an issue, it is essential
to look at the lie (L) scores which are incorporated in the EPQ. High P scorers
can be taken at face value, but low P scorers may well be dissembling, there
usually being a PL correlation of around —0.4. To obtain a clearer picture it
would be wise to look at the L scores of low P scorers, querying the veracity of
high L low P scorers.
Similarly, an extraverted high P scorer is likely to have different character-
istics to an introverted one. High PE people may be very adventurous, risk
taking, impulsive, fun loving, and sensation seeking. But when the concept of N
is introduced, the picture changes. High PE people with high N scores may be
120 Personality

antisocial, not accepting social mores, but the N, if elevated enough, can act
rather like a brake on dangerous, foolhardy sensation seeking, making would-
be adventurers a little more cautious.
The characteristic profile of hard drug addicts is high PN and low EL. In
other words, the pathological scores are elevated and the introverted low lie
scores add to the "I don't care what anybody thinks" attitude.
Finally, high P in very intelligent people may be infinitely more dangerous
when the person concerned is psychopathic, whereas a dull high P scorer might
be easier to capture and stop in his tracks. Conversely, a bright high P scorer
who is gifted and creative can be responsible for many valuable inventions or
works of art, while dull high P scorers would be unlikely to achieve anything
worthwhile.
Consequently, we are concerned when certain studies look at any of the
dimensions in isolation. They are not getting the true picture, and this is
particularly so when ignoring the lie scores.

REFERENCES

Allsopp, J. F. & Feldman, M. P. (1974). Extraversion, neuroticism, psychoticism and

antisocial behaviour in school girls. Social Behaviour and Personality, 2, 184-190.
Allsopp, J. F. & Feldman, M. P. (1976). Item analysis of questionnaire measures of
personality and antisocial behaviour in school boys. British Journal of Criminology, 16,
337-351.
Eaves, L. J., Eysenck, H. J., & Martin, N. G. (1989). Genes, culture and personality: an
empirical approach. New York: Academic Press.
Eysenck, H. J. (1950). Criterion analysis—an application of the hypothetico-deductive
method to factor analysis. Psychological Review, 57, 38-53.
Eysenck, H. J. (1952). The scientific study of personality. London: Routledge & Kegan
Paul.
Eysenck, H. J. (1953). On criterion analysis. A reply to F. W. Beezhold. Journal of
National Institute of Personnel Research, Johannesburg, 5, 183-187.
Eysenck, H. J. (1956). Maudsley Medical Questionnaire. The questionnaire measure-
ment of neuroticism and extraversion. Revista di Psicologia, 50, 113-140.
Eysenck, H. J. (1959). The Maudsley Personality Inventory. London: University of
London Press.
Eysenck, H. J. (1991). Dimensions of personality: 16, 5 or 3?—Criteria for a taxonomic
paradigm. Personality and Individual Differences, 12, (8), 773-790.
Eysenck, H. J. (1992). The definition and measurement of psychoticism. Personality and
Individual Differences, 13, 757-785.
Eysenck, H. J. (1995). Genius. The natural history of creativity. Cambridge University
Press, Cambridge.
Eysenck, H. J. (in press). Addiction, personality and motivation. Psychopharmacology.
Eysenck, H. J., & Eysenck, S. B. G. (1964). Manual of the Eysenck Personality Inventory.
London: University of London Press.
 Psychoticism as a dimension of personality 121

Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality

Questionnaire. London: Hodder & Stoughton.
Eysenck, H. J., & Eysenck, S. B. G. (1976). Psychoticism as a dimension of personality.
London: Hodder & Stoughton.
Eysenck, H. J., & Eysenck, S. B. G. (1991). Eysenck Personality Scales (EPS Adult).
London: Hodder & Stoughton.
Eysenck, H. J., Granger, G. W., & Brengelmann, J. C. (1957). Perceptual processes and
mental illness. London: Oxford University Press.
Eysenck, S. B. G. (1981). Impulsiveness and antisocial behaviour in children. Current
Psychological Research, 1, 31-37.
Eysenck, S. B. G., Easting, G., & Pearson, P. R. (1984). Age norms for impulsiveness,
venturesomeness and empathy in Children. Personality and Individual Differences, 5,
(3), 315-321.
Eysenck, S. B. G., & Eysenck, H. J. (1978). Impulsiveness and venturesomeness: their
position in a dimensional system of personality description. Psychological Reports, 43,
1247-1255.
Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. P. (1985). Age norms for
impulsiveness, venturesomeness and empathy in adults. Personality and Individual
Differences, 6, (5), 613-619.
Feldman, R., & Eysenck, S. B. G. (1986). Addictive personality traits in bulimic
patients. Personality and Individual Differences, 7, 923-926.
Gossop, M., & Eysenck, S. B. G. (1980). A further investigation into the personality of
drug addicts in treatment. British Journal of Addiction, 75, 305-311.
Gossop, M., & Eysenck, S. B. G. (1983). A comparison of the personality of drug
addicts in treatment with that of a prison population. Personality and Individual
Differences, 4, 207-209.
Gray, N., Pickering, A., & Gray, J. (1994). Psychoticism and dopamine D2 binding in
the basal ganglia using single photon emission tomography. Personality and Individual
Differences, 17, (3), 431-434.
Hare, R. D. (1980). A research scale for the assessment of psychopathy in criminal
populations. Personality and Individual Differences, 1, 111-120.
Lane, D. (1977). Persistent failure and potential success: a study of failure to respond to
therapy in children. Unpublished M. Phil. Thesis, University of London.
Mehrabian, A., & Epstein, N. (1972). A measure of emotional empathy. Journal of
Personality, 40, 525-543.
Ogden, M. E., Dundas, M., & Bhat, A. V. (1988). Personality differences among
alcoholic misusers in community treatment. Personality and Individual Differences, 10,
(2), 265-267.
Rahman, M. A., & Eysenck, S. B. G. (1978). Psychoticism and response to treatment in
neurotic patients. Behaviour Research and Therapy, 16, 183-189.
Teasdale, J., Seagraves, R., & Zacune, J. (1971). Psychoticism in drug users. British
Journal of Social and Clinical Psychology, 10, 160-171.
Zuckerman, M. (1991). Psychobiology of personality. Cambridge: Cambridge University
Press.
Chapter 7
Classical conditioning, arousal, and crime:
A biosocial perspective

A. Raine

1. INTRODUCTION

The pleasure in contributing to the Festschrift of one of the world's leading

psychologists is particularly heightened by the fact that Hans Eysenck played a
major role in my very first publication. This first publication in turn has shaped
my entire research career, and the research focus I have today (the biosocial
bases of crime and violence) is a direct result of this publication. The
publication was in itself based on the very first public talk I gave, a talk that I
shall never forget. It took place at the British Psychophysiology Society
meeting at the Institute of Psychiatry, London, in December 1980 while I was a
graduate student under Peter Venables (himself a former graduate student of
Hans Eysenck) at York University.
The talk was a source of immense anxiety for me not only because it was my
very first public speech, but also because my results failed to support Eysenck's
primary prediction (that antisocials condition poorly), and instead indicated a
biosocial interaction such that antisocials from good homes are bad
conditioners, whilst antisocials from bad homes are paradoxically good
conditioners. My concern was that I would be totally humiliated by a deluge
of criticism from the master theorist on his home ground. I felt somewhat
better when I looked round just before my talk and could see no sign of
Professor Eysenck, and was most relieved to stumble through the talk and
follow-up questions from non-Eysenckians. Imagine my horror however when,
walking from the podium up the steps of the main auditorium in the Institute
of Psychiatry for the afternoon tea-break, I saw Professor Eysenck lying in wait
for me at the top of the stairs. Imagine my shock when, having reached the top
 Classical conditioning, arousal, and crime: A biosocial perspective 123

of the stairs, he introduced himself, congratulated me on the paper, and

suggested that I might consider submitting it to Personality and Individual
Differences for potential publication!
It is hard to convey what a profoundly exhilarating boost this very first
meeting with Hans Eysenck gave to a young unpublished student greatly
lacking in confidence, and how much inspiration it gave to my future work.
Nevertheless, what I hope I will be able to convey in this chapter is an academic
sense of what Eysenck's theorizing on crime and conditioning has been, to
what extent his theory has received empirical support, how Eysenck has
advocated a more sophisticated biosocial perspective on understanding crime,
and where this approach will take us in the future.
In doing so I hope to convey some of the brilliance and genius behind
Eysenck's thinking, which was very much ahead of its time. I hope nevertheless
that the more important lesson which I learnt from Hans Eysenck on that day,
that of senior scientists giving encouragement to young researchers, is not
forgotten in the process. To my mind, the extent of Eysenck's contributions lie
not just with his own work, but also with the larger scientific corpus that he has
stimulated in others with his own generous encouragement.

2. EYSENCK'S CONDITIONING THEORY OF CRIME

Eysenck's influential theory of criminal behavior rests on the notion that

criminals and other antisocials are deficient with respect to classical
conditioning (Eysenck, 1964; 1977; Eysenck & Gudjonsson, 1989). He argued
that classical conditioning is fundamental to the whole process of socialization
whereby the individual learns to withhold antisocial responses. It is argued that
the crucial mechanism that stops most of us from committing criminal and
antisocial acts is the concept of conscience; a well-developed conscience is what
holds many of us back from not stealing even in those situations when we are
almost certain of getting away with the theft undetected. Eysenck argues that
what we call "conscience" is, in effect, a set of classically-conditioned
emotional responses. The greater the individual's ability to develop and form
classically-conditioned emotional responses, the greater the conscience
development, and the less likely will be the probability of becoming antisocial.
Conversely, poor conditionability will result in poor conscience development
and undersocialized, antisocial behavior.
Classical conditioning involves learning that an initially neutral event (a
conditional stimulus, CS), when closely followed in time by an aversive event
(unconditional stimulus, UCS), will develop the properties of this UCS. In the
classic case of Pavlov's dogs, a bell (CS) was paired with the presentation of
food (UCS). Food to hungry dogs automatically elicits an unconditional
response (UCR), in this case salivation. After a sufficient number of pairings of
124 Personality

the bell with the food, the bell by itself will come to elicit the UCR—or
salivation. Although conditioning has often been viewed as reflecting auto-
matic, reflexive processes, experiments in human autonomic classical con-
ditioning support the notion that complex cognitive processes are involved in
this form of learning (see e.g., Dawson & Scheil, 1985).
In concrete terms, the way classical conditioning is hypothesized by Eysenck
to relate to socialization is as follows. Taking the scenario of a small child
stealing a cookie (CS) from the kitchen, punishment by the parent (scolding or
physical punishment—UCS) elicits an unconditional response (UCR) whereby
the child is upset and feels uncomfortable. After a number of similar "learning
trials," the sight of the cookie (or even the thought of stealing the cookie) will
elicit an uncomfortable feeling in the child (conditional response—CR) which
acts to avert the child from enacting the "theft." Similar "conditioned
emotional responses" developed relatively early in life in varying situations
combine, in Eysenck's view, to represent what we call "conscience."
In this analysis, socialized individuals develop a feeling of uneasiness at even
contemplating a criminal act (robbery, assault) presumably because such
thoughts elicit representations or "unconscious" memories of punishment
early in life for milder but related misdemeanors (theft, behaving aggressively).
In this context the common response of socialized individuals to crimes
committed by others such as "I could never even think of doing such a thing"
becomes understandable. Socialized individuals do not even contemplate such
events because even the thought of such acts elicits CRs involving discomfort.
Eysenck's theory of crime involves more concepts than just conditioning. He
argues that crime has a genetic basis, and that genetic differences lead to
individual differences in CNS and ANS functioning which in turn shape
personality and behavior. Central to such differences are individual differences
in arousal. These differences in arousal levels result in differing degrees of both
extraversion and conditionability, with low levels of arousal predisposing to
poor conditionability and high levels of extraversion. Because of links between
extraversion and both arousal and conditionability, Eysenck (1964, 1977) went
on to argue that criminals would be extravert, and also developed predictions
concerning high neuroticism and psychoticism (H. J. Eysenck & S. B. G.
Eysenck, 1978; H. J. Eysenck, 1987).

2.1 Evidence for poor classical conditioning in antisocials

Eysenck's predictions concerning personality have generated a great deal of
research, and reviews of findings may be found in Bartol (1991), Eysenck
(1987, in press), Farrington, Biron, and LeBlanc (1982), and Passingham
(1972). In general, the findings support the predictions, particularly with
respect to Neuroticism in adult samples, Extraversion in juvenile samples, and
Psychoticism in both adult and juvenile samples (Eysenck, in press).
 Classical conditioning, arousal, and crime: A biosocial perspective 125

Nevertheless, the central idea in Eysenck's theory is that criminals and

antisocials will be characterized by poor classical conditioning. This prediction
has received less empirical attention, probably because it is much easier to
assess personality than conditionability. Classical conditioning has most fre-
quently been assessed using skin conductance: a neutral tone (CS) is presented
to the subject, followed a few seconds later by either a loud tone or an electric
shock (UCS). The key measure derived from this paradigm is the size of the
skin conductance (SC) response elicited by the CS after a number of CS-UCS
pairings. The lower the SC amplitude, the poorer the degree of conditioning.
On occasions however, eye-blink classical conditionability has been assessed in
which a neutral stimulus is followed by an air-puff to the eye which elicits an
eye-blink (UCR)—the measure here is the magnitude of the eye-blink to the
CS alone after a number of CS-UCS pairings.
The last systematic review of conditionability was reported by Hare (1978).
Of the 14 studies reported by Hare covering classical conditioning and what is
termed "quasi-conditioning" (see below), 12 indicated that psychopaths,
criminals, delinquents, and antisocials showed poorer SC conditioning than
control groups. In one of the remaining two studies, significantly poorer con-
ditioning was observed for a subgroup of psychopaths (those with low scores on
the Socialization scale). The remaining study failed to observe overall
significant effects, and instead observed that younger psychopaths gave larger
responses than older psychopaths. Overall, therefore, this review indicates
general support for the notion of poorer conditionability in antisocial groups.
In order to assess whether this general conclusion remains true, findings
from conditioning studies conducted since 1978 have been assessed by Raine
(1993). These six studies and their key findings are noted in Table 7.1. SC
conditioning in these studies is assessed either by SCRs occurring to the
conditional stimulus (what has been termed the conditioned "A" response) or
by the SCR occurring in between the CS and the unconditional stimulus (the
conditioned "B" response). In two of the studies (Hare, 1982; Tharp,

Table 7.1. Key findings from studies since 1977 on classical conditioning in antisocials, criminals,
and psychopaths as measured by skin conductance

Authors' significant findings Subjects Finding

Ziskind et al. (1978) Psychopathic gamblers Poor differential conditioning
but verbal awareness
Aniskiewicz et al. (1979) Primary psychopaths Poor vicarious conditioning
Tharp et al. (1980) Psychopathic gamblers Less anticipatory responding
Raine and Venables (1981) Conduct disorder Poor conditioning in high social
class antisocials
Hemming (1981) Criminals from good homes Less conditioned discrimination
in extinction
Hare (1982) Psychopaths Less anticipatory responding
126 Personality

Maltzman, Syndulko, & Ziskind, 1980) the paradigm consists of a "count-

down" procedure in which the subject awaits the onset of an aversive stimulus
whose onset is signaled several seconds beforehand—a paradigm referred to by
Hare (1978) as "quasi-conditioning." All six of these studies showed some
evidence indicating significantly poorer SC conditionability in antisocials. Not
all of these studies provide unequivocal evidence for poor conditioning
however. Hemming (1981) found group differences for conditioned dis-
crimination in extinction, but not for conditioning discrimination during
acquisition. Similarly, Raine and Venables (1981) found poor conditioning
specifically in antisocial children from higher social class, but not in those from
lower social classes, an issue that will be referred to in more detail later.

2.2 Assessment of conditioning studies

The findings outlined in Table 7.1 are unusual in that, in one way or another,
they all find significant group differences even though there are wide variations
in these studies. For example, paradigms varied from a classical CS-UCS
paradigm (e.g., Hemming, 1981) to vicarious conditioning where subjects
watching others receive electric shocks following a CS (Aniskiewicz et al.,
1979) to quasi-conditioning count-down procedures (e.g., Tharp et al., 1980).
Subjects ranged from uninstitutionalized antisocial children (e.g., Raine &
Venables, 1981) to adult criminals (Hemming, 1981) to institutionalized
psychopaths (Tharp et al., 1980) to psychopathic gamblers (Ziskind, Syndulko,
& Maltzman, 1978). The fact that all studies showed significant effects in the
predicted direction would indicate that poor conditioning is related to the
general development of antisocial behavior.
Several of these paradigms do not control for factors such as sensitization.
However, it is likely that the SC conditioning measures obtained are a strong
correlate of true SC conditioning, since one would expect that those who
sensitize easily also condition easily.
There are several interesting aspects to some of these studies. Ziskind et al.
(1978) demonstrated that while psychopaths were able to verbalize the
contingency between the CS and UCS (i.e. they know that the warning tone
was followed by the aversive tone), they did not show conditioning. This
finding suggests that conditioning deficits in antisocials are not merely a
reflection of a cognitive, conscious process involving understanding the link
between the CS and UCS, but may involve more deep-seated, "unconscious"
or pre-attentive processes.
The study by Hemming (1981) is of interest in that the subject population
consisted of criminals from relatively good social backgrounds. It has
previously been argued that biological predispositional variables may have
greater explanatory power in antisocials from relatively benign homes where
the "social push" towards antisocial behavior is low; if individuals become
 Classical conditioning, arousal, and crime: A biosocial perspective 127

antisocial therefore, it may be more for biological than social reasons

(Mednick, 1977; Raine & Venables, 1981). Hemming's findings are consistent
with this analysis. An early finding by Lykken (1955) also appears to be
consistent with this approach. Lykken observed that primary psychopathic
inmates showed poorer SC conditioning to an electric shock than neurotic
psychopaths. In commenting on Lykken's subject selection procedures, Siddle
and Trasler (1981) point out that in this study subjects were excluded if they
came from a "markedly sociopathic or deviant" family background (Lykken,
1955, p. Ilia). As such, SC conditioning deficits were found in psychopaths
from relative good home backgrounds. This issue is an important one and it
leads to the notion that the social environment may mediate antisocial-
conditioning relationships.

3. EYSENCK'S BIOSOCIAL THEORY OF CRIME AND CONDITIONING

Throughout his career, Eysenck has been repeatedly and mistakenly accused of
being a radical biotrope who rigidly espouses a genetic and biological approach
to human behavior. Paradoxically, quite the opposite has been the case.
Eysenck has always acknowledged the important role of the environment in
shaping human behavior, while at the same time making it clear that genetics
and biology do play a significant role. This perspective is especially true of
Eysenck's approach to crime. Indeed, he viewed the interaction between
environmental and biological factors as absolutely critical to the development
of crime.
Surprisingly, this biosocial perspective has not received the attention that it
deserves, yet Eysenck made his biosocial perspective quite clear. In discussing
the hypothetical case of a child who, instead of having law-abiding parents, has
a mother who is a prostitute and a father who is a thief:
Clearly the exact opposite of what we have posited heretofore will take place. Now it
will be the introverted child, the child who conditions well, who will condition to the
precepts emerging from this "Fagin's kitchen." Instead of becoming conditioned to
be a good and law-abiding citizen, we now have our introvert being conditioned to be
a "good" law-breaking thief or prostitute. (Eysenck, 1977, pp. 150-151)
Eysenck (1977) therefore argued that children who are highly conditionable
and who have antisocial parents will become "socialized" into their parents'
antisocial habits, whereas children who condition poorly will, at least in this
environment, paradoxically avoid becoming antisocial.
Eysenck's biosocial prediction was tested by Raine and Venables (1981). In
this study 101 schoolboys were assessed on skin conductance conditioning
while their antisocial behavior was assessed by (1) teacher ratings of antisocial
school behavior (Quay & Parsons, 1970), and (2) a factor of self-report
128 Personality

antisocial personality with high loadings from several antisocial scales such as
Socialization (—.72 loading), Unsocialized-Psychopathic (.79), as well loadings
from personality variables such as Psychoticism (.62), Impulsivity (.59), and
Neuroticism (.58). The conditioning paradigm was designed to test Eysenck's
theory and features included partial reinforcement, a relatively weak UCS, and
short interstimulus interval which are viewed as favoring introverts (see Raine
& Venables for more details). Social class was used to assess quality of home
environment, with low social class being a proxy for a relatively more crimino-
genic home environment.
The expectation stemming from Eysenck's theory would be that antisocials
from good home environments would show the expected conditioning deficit,
while antisocials from bad home environments would show good condition-
ability. This conclusion is indicated by the fact that antisocial measures
correlated significantly and negatively with conditioning in the high social class
group, but in the positive direction in the low social class group.
To illustrate the findings graphically, subjects were dichotomized into
"antisocial" and "prosocial" groups on the basis of a median split on the
antisocial measures. Results of this analysis are illustrated in Figure 7.1. A
significant interaction was observed between social class and antisocial
grouping in relation to conditioning (p < .04). As illustrated in Figure 7.1,
Eysenck's predictions were supported. Antisocials from benign homes showed

0.07 Antisocials
Prosocials

0.06

0.05

0.04

O 0.03
to

0.02
D

0.01

Low class High class

Figure 7.1. The interaction between social class and antisocial behavior in relation to skin
conductance classical conditionability (Raine & Venables, 1981).
 Classical conditioning, arousal, and crime: A biosocial perspective 129

poor conditioning, while antisocials from negative home backgrounds showed

good conditionability. Although it appears that the effect is stronger in low
than high social classes, the correlational analyses indicated equal strength of
findings in the two social class groups.
Poorer SC conditioning was observed in children from higher social classes.
One finding from this study which does not fit so easily with this perspective is
that antisocials from lower social classes showed relatively good conditioning
(see Figure 7.1). This specific finding may be more easily explained by the
process of "antisocialization." If low social class indirectly reflects a relatively
criminogenic environment, then the stronger conditionability in children from
low social classes found in Raine and Venables (1981) would be consistent with
this analysis.
Although this study supports Eysenck's biosocial theory of crime, there are
clear limitations. First and foremost, low social class is at best an indirect
measure of criminogenicity of the home environment. Better studies which
more directly assess antisocial processes in subject's homes and peer groups in
conjunction with conditionability are needed to confirm Eysenck's hypothesis.
Nevertheless, the important point to make is that conditionability may interact
with social factors in important ways to explain antisocial behavior, a fact which
provides some encouragement for a biosocial perspective on crime.

4. NEW FINDINGS SUPPORTING THE BIOSOCIAL PERSPECTIVE

New findings which provide some additional support for Eysenck's biosocial
perspective are derived from initial analyses of the Mauritius longitudinal
study. In assessing these findings, it is important to bear in mind that they
constitute provisional and initial analyses. The dependent variable in this case
is not conditionability per se but skin conductance orienting, a measure of
conceptual relevance to conditioning (see below). In brief, subjects consist of
1795 male and female children who were psychophysiologically tested at age 3
years (Venables, 1978). 51% are male and 49% are female. The two main
ethnic groups consist of Indian (69%) and Creole (29%).
All 1795 subjects were assessed on SC orienting and resting heart rate at age
3 years (see Venables, 1978 for full details). Inhibited versus disinhibited
temperament was assessed at age 3 years (see Scarpa, Raine, Venables, &
Mednick, 1995 for full details). Out of the total of 1795,1213 were assessed by
teachers at age 11 years on the Achenbach scale (Achenbach & Edelbrock,
1979). Analyses below focus on two key subscales of this checklist, Aggression
and Delinquency. Two limitations of the Achenbach measures are that (1)
scales for males and females are somewhat different, and (2) the "Aggression"
scale contains many items with no aggression component, while the
"delinquency" scale contains aggression items. To provide purer indices of
130 Personality

Aggression and Non-Aggressive Delinquency subscales common to both sexes,

two new scales were constructed. Coefficient alpha for the Aggression scale
were 0.72 (boys) and 0.72 (girls), with slightly lower reliabilities found for
Delinquency (0.64 for boys, 0.68 for girls).
The key psychosocial variable to be considered in the analyses below was
socioeconomic status (SES). This was taken at age 3 years, and consisted of a
factor score based on a factor analysis of a variety of social variables which
produced one major factor. Variables loading on this factor included number
of years of education of the parents, parental occupation, additional
educational training of the parents, appearance of the home, number of
rooms per person, and number of rooms in the house. Data were available on
1321 of the subjects. Upper and lower quartile splits were used to divide
subjects into high and low Aggression, Delinquency, total Antisocial, and
Disinhibited groups. High and low SES groups were then formed on the basis
of a median split.
The ANOVA on frequency of SC orienting responses produced a significant
Aggression SES interaction, F(1183) = 6.5, p < .01. The interaction is illus-
trated in Figure 7.2. It can be seen that in the high SES group, Aggressives tend
to give fewer orienting responses than Non-Aggressives, whereas this effect is
reversed in the lower SES group where Aggressives showed greater orienting.
As such, the effect is in the same direction as for the conditioning SES
interaction observed by Raine and Venables (1981).

High aggression
Low aggression

2.5

1.5

0.5

Lowses High ses

Figure 7.2. The interaction between social class and aggression in relation to skin conductance
orienting (Raine et al., in press).
 Classical conditioning, arousal, and crime: A biosocial perspective 131

This biosocial effect was not specific to aggressive behavior as the same
interaction effect was observed for nonaggressive Delinquency (F(1125) = 4.3,
p < .04), with the same pattern of results emerging. No interactions were
observed with either sex or ethnicity.

4.1 Interpretation of new findings

The biosocial effect for orienting is of significance in that it mirrors the effect
observed by Raine and Venables (1981) for SC conditioning. It takes that
previous support for Eysenck's biosocial theory four steps further by showing
(1) orienting and SES prospectively collected at age 3 years predicts to
aggression at age 11 years, (2) the effects for males also hold for females, (3)
the effects generalize across ethnic groups, and (4) the effects appear to apply
to delinquency as well as aggression.
These orienting results can be interpreted in the way that Raine and
Venables (1981) interpreted their conditioning data along the lines of
Eysenck's biosocial theory, that is, in terms of the benign homes effect (poor
conditioning characterizes antisocials from benign home backgrounds) and the
antisocialization effect (good orienting characterizes antisocials from poor
home backgrounds). SC orienting is a sensitive measure of information
processing (Dawson, 1990; Dawson, Schell, & Filion, 1991). Poor orienting is
thought to reflect a fundamental deficit in the ability to allocate attentional
resources to environmental events. As such, poor orienting in antisocials from
benign homes may reflect an attentional deficit which could be expected to
retard classical conditioning and the ability to form associations between
signals of punishment and the punishment itself. Good orienting in aggressives
from poor homes may reflect good attention and more proficient learning of
antisocial habits in more criminogenic homes.

4.2 Extensions of Eysenck's biosocial theory to brain imaging findings

Increased SC orienting has been associated with better prefrontal functioning
(Hazlett et al., 1993) and increased area of the prefrontal cortex (Raine,
Reynolds, & Sheard, 1991). Furthermore, classical conditioning is associated
with increased cerebral blood flow in the prefrontal cortex (Hugdahl et al.,
1995). As such, poorer orienting and conditioning are associated with
prefrontal deficits. Because antisocials from good home backgrounds have
been shown above to have poor orienting and poor conditioning, we might
expect these individuals to represent a subgroup of violent criminals who are
particularly characterized by prefrontal deficits. Conversely, antisocials from
negative home backgrounds would not be expected to show these deficits, or
may even have good prefrontal functioning.
132 Personality

We have recently tested this prediction using data from a brain imaging
study of murderers (Stoddard, in press). We had previously found that
murderers, compared to age and sex-matched normal controls, have selective
reductions in glucose metabolism in the prefrontal region of the brain. Glucose
metabolism was assessed using positron emission tomography (PET) and using
the continuous performance task to challenge the prefrontal cortex. To test the
above biosocial hypothesis, murderers were divided into those with and
without a home background characterized by psychosocial deprivation (e.g.,
physical and sexual abuse, neglect, extreme poverty, severe family conflict).
Results of the analysis are shown in Figure 7.3. It can be seen that the lowest
prefrontal functioning was observed in murderers who lacked psychosocial
dysfunction. This group had left prefrontal metabolic rates which were
significantly lower than both controls and murderers with psychosocial deficits
(p < .05), and right prefrontal rates that were significantly lower than controls.
These findings are consistent with Eysenck's theory that would predict that
criminals from benign home backgrounds are most likely to exhibit biological
deficits. However, the biosocial theory is not fully supported because

Figure 7.3. Low prefrontal glucose metabolism in murderers who lack psychosocial deficits
(Stoddard, in press).
 Classical conditioning, arousal, and crime: A biosocial perspective 133

murderers with psychosocial deficits did not have significantly higher

prefrontal values than controls. This may be because the sample is weighted
towards those with negative home backgrounds and lacking violent offenders
from completely trouble-free homes. Despite this caveat, extrapolating from
conditioning to orienting to prefrontal functioning does provide some limited
support for Eysenck's biosocial hypothesis.

4.3 Underarousal and antisocial behavior

The focus of this chapter has been on classical conditioning because it is the
crucial process in Eysenck's theory of crime. Mention should also be briefly
made of the other psychophysiological process of underarousal. Eysenck
(1977) also invoked this construct because he believed that low arousal is
associated with extraversion, poorer conditioning, and hence antisocial
behavior. There has been surprisingly good support for this prediction,
particularly with respect to autonomic arousal (see Raine, 1993, for a full
review). Particularly important are positive findings from three prospective
longitudinal studies because they allow for temporal ordering of variables, and
hence a better test of causality.
One problem with many of these studies, both cross-sectional and
longitudinal, is that evidence is based on only one measure of arousal. A
nine-year prospective study of crime by Raine, Venables, and Williams (1990)
has shown, however, that HRL, NSFs, and excessive theta EEG measured at
age 15 years in normal unselected schoolboys predicted criminal behavior at
age 24 years. These three measures correctly classified 74.7% of all subjects as
criminal/noncriminal, a rate significantly greater than chance (50%). In the
total population the three arousal measures were statistically independent; the
fact that they all independently predicted criminal behavior indicates strong
support for an arousal theory of criminal and antisocial behavior (although this
finding also cautions against the use of a simplistic, unitary arousal concept in
explaining crime). Group differences in social class, academic ability, and area
of residence were not found to mediate the link between underarousal and
antisocial behavior. This is the first study providing evidence for Eysenck's
underarousal perspective of crime which uses all three psychophysiological
response systems.

4.4 Psychophysiological protective factors against crime development

One of the hallmark's of Eysenck's theoretical contribution to psychology is
that he has the unusual creative ability to pose simple, powerful questions
which are rarely asked. For example, rather than asking the question "why do
children become antisocial," he has asked "why don't all children become
antisocial"? This simple yet challenging question provided the basis of
134 Personality

Eysenck's conditioning and personality theory of socialization and crime

(Eysenck, 1977). All psychophysiological research to date has attempted to ask
the question "what psychophysiological factors predispose to crime?," and
consequently has focused exclusively on risk factors for crime development.
Following Eysenck's lead, a potentially more important question to be posed,
however, is "what factors protect a child predisposed to crime from becoming
criminal?" Understanding biological protective factors against crime develop-
ment may be of critical conceptual importance because it can more directly
inform intervention and prevention of antisocial behavior. Though this seems
an obvious line of thinking, it has not been pursued in biological research on
crime until very recently. These new data provide support for what Eysenck
himself might have predicted, that is, high arousal, orienting, and condition-
ability characterize those who desist from crime.
Raine, Venables, and Williams (1995) report on a 14-year prospective study
in which autonomic and CNS measures of arousal, orienting, and classical
conditioning were taken in 101 unselected 15-year-old male schoolchildren. Of
these, 17 adolescent antisocials who desisted from adult crime (Desistors) were
matched on adolescent antisocial behavior and demographic variables with 17
adolescent antisocials who had become criminal by age 29 (Criminals) and 17
nonantisocial, noncriminals (Controls). Desistors had significantly higher heart
rate levels, higher SC arousal (measured by nonspecific SC responses—see
Figure 7.4), and higher SC orienting, better SC conditioning, and faster half-
recovery time of the SC response (thought to reflect an open attentional stance
to the environment) relative to Criminals (see Figure 7.5). Findings suggest
that individuals predisposed to adult crime by virtue of showing antisocial
behavior in adolescence may be protected from crime by heightened levels of
autonomic arousal and reactivity.
Good conditioning and fast fear dissipation/open attentional stance may
protect against criminal behavior because they help facilitate the development
of learning processes (specifically, classical conditioning and passive avoidance
learning) which have been theoretically viewed by Eysenck as underpinning the
process of socialization (Eysenck, 1977). Such an advantageous psychophysio-
logical profile does not, however, explain why Desistors were antisocial in
adolescence. It seems feasible that this subgroup were predisposed to
antisocial behavior for more transient, nonbiological reasons, such as negative
peer influences (Blumstein, Cohen, Roth, & Visher, 1986), which may not
carry over into adulthood. Developing further Eysenck's biosocial perspective
on crime, emphasizing an interaction between biological processes and social
processes, may nevertheless yield some clues. For example, Moffitt (1993) has
argued that antisocial behavior during adolescence is actually normative social
behavior arising as a response to the contemporary secular context. It is
conceivable that good conditioners are well-behaved in the prevailing prosocial
environments they experience in early development, but may for a temporary
 Classical conditioning, arousal, and crime: A biosocial perspective 135

Figure 7.4. High resting (a) heart rate and (b) skin conductance characterized antisocial
adolescents who desist from crime in adulthood (Raine et al., 1995).
136 Personality

Figure 7.5. Better classical conditioning and orienting, and faster half-recovery times in antisocial
adolescents who desist from crime in adulthood (Raine, Venables, & Williams, 19%).
 Classical conditioning, arousal, and crime: A biosocial perspective 137

period become easily conditioned into the antisocial mores that predominate
only during adolescence (Moffitt). A change back to prosocial behavior may
occur as the good conditioner leaves these antisocial peers and becomes
influenced by a different set of reinforcement contingencies and a more
prosocial life norm (e.g., starting work, marriage, having children, setting up a
home). Moffitt has argued that adolescent-limited antisocial behavior may be
more under the control of reinforcement and punishment contingencies;
heightened classical conditionablility in the Desistor group may in turn make
this adolescent subgroup particularly susceptible to the prevailing reinforce-
ment schedules.

5. IMPLICATIONS OF EYSENCK'S EARLY THEORIZING FOR A NEW

GENERATION OF CRIME RESEARCH

Eysenck was decades ahead of his time in suggesting a biosocial approach to

crime, for it is only now that this approach is beginning to be embraced by a
wider scientific community. In what ways can future research build on the solid
foundation laid down by Eysenck? If Eysenck was starting his career again in
1997, what would his blueprint be for tackling the growing problem of crime
and violence in society?
At one level I suspect several features of his theorizing would remain intact.
He would still argue that it is critical to discover the genetic and biological
underpinnings to crime and violence, and part of this would still involve key
aspect of his personality theory. He would still emphasize a biosocial approach
which attempts to integrate these individual difference trait variables with
social and situational influences. He would still argue that we need to apply
what we have learnt from scientific inquiry to tackle crime in society directly.
While these issues may remain fundamental, one suspects there may be both
a theoretical and methodological shift in his approach. In terms of
methodology, he might argue for a molecular genetic approach to furthering
our understanding of the basic biological, temperamental, and personality
predispositions to crime. He might advocate a discordant twin approach to
attempting to understand what environmental factors help protect a mono-
zygotic twin genetically predisposed to crime (by virtue of the co-twin being
criminal) from becoming criminal.
The recent technical advances brought about by brain imaging would
certainly result in suggestions about using functional magnetic resonance
imaging (MRI) to assess arousal and conditionability more directly. The
increase in our knowledge of brain functioning might have led Eysenck to
speculate more on dysfunction to specific brain mechanisms which may
underlie deficits in arousal, conditioning, and emotion regulation, and the
neural networks that subserve antisocial and aggressive behavior. For example,
138 Personality

the prefrontal cortex is involved in the regulation of arousal, and dysfunction to

this structure has been implicated in violent offenders (Raine et al., 1994).
Nevertheless, one suspects that he would still advocate the use of autonomic
psychophysiology to understanding crime, because while bran imaging tech-
niques are excellent tools to understand arousal and cognition in the CNS, skin
conductance and heart rate are still state-of-the-art measures for obtaining a
handle on autonomic functions which must be central to any theory of crime
which focuses on emotional responding.
At a conceptual level one suspects that there would be a greater focus on the
role of Psychoticism in his personality theory of crime. Again, Eysenck was ahead
of his time in developing a scale which assesses the interface between
schizophrenia/psychoticism and crime. Linking crime with schizophrenia has
been strongly resisted for decades, and it is only very recently that the larger
scientific community have accepted a link between these conditions. It may be
that criminals with schizotypal-like features differ in terms of underlying etiology
relative to criminals lacking such characteristics, and future research may usefully
explore further the contribution of Psychoticism to the etiology of crime.
Most importantly of all, I suspect Eysenck would develop further his notions
on the biosocial bases of crime. Not only does the social environment moderate
the relationship between conditioning/arousal and antisocial behavior, but also
it is quite possible that early environmental processes can lead to changes in
autonomic functioning which may then predispose to crime along the lines
suggested by Eysenck. For example, Wadsworth (1976) showed that children
who come from homes broken by divorce or separation before the age of 4
years are more likely to have low heart rates at age 11 years, while those who
have high heart rates at age 11 years are more likely to become violent criminal
offenders in early adulthood. Such environmental influences on biological
influences may give important pointers for future intervention and prevention
research.
These are some of the possible ways that one imagines a born-again Hans
Eysenck would reshape the field of crime research. Yet these are just guesses,
and clearly there are more future developments that stem from Eysenck's
seminal theory of crime. Perhaps any additional speculations are best left to
Hans Eysenck himself!

REFERENCES

Achenbach T. M. & Edelbrock C. S. (1979). The Child Behavior Profile: II. Boys aged
12-16 and girls aged 6-11 and 12-16. Journal of Consulting and Clinical Psychology,
47, 223-233.
Aniskiewicz, A. (1973). Autonomic components of vicarious conditioning and
psychopathy. Journal of Clinical Psychology, 35, 60-67.
 Classical conditioning, arousal, and crime: A biosocial perspective 139

Bartol, C. R. (1991). Criminal behavior: A psychosocial approach (3rd ed.). New Jersey:
Prentice Hall.
Blumstein, A., Cohen, J., Roth, J., & Visher, C. (1986). Criminal careers and "career
criminals." Washington, DC: National Academy Press.
Dawson, M. E. (1990). Psychophysiology at the interface of clinical science, cognitive
science, and neuroscience: Presidential address, 1989. Psychophysiology, 27, 243-255.
Dawson, M. E., Filion, D. L., & Schell, A. M. (1989). Is elicitation of the autonomic
orienting response associated with allocation of processing resources? Psychophysiol-
ogy, 26, 560-572.
Dawson, M. E., & Schell, A. (1985). Information processing and human autonomic
classical conditioning. In P. K. Ackles, J. R. Jennings, & M. G. H. Coles (Eds.)
Advances in psychophysiology (Vol. 1, pp. 89-165). Greenwich, CT: JAI Press.
Dawson, M. E., Schell, A., & Filion, D. (1990). The electrodermal system. In J. T.
Cacioppo, & L. G. Tassinary (Eds.). Principles of psychophysiology (pp. 295-324).
Cambridge: Cambridge University Press.
Dawson, M. E., Schell, A. M., & Filion, D. (1991). The electrodermal system. In J. T.,
Cacioppo, L. G. Tassinary, & A. Fridlund (Eds). Principles of psychophysiology:
Physical, social and inferential elements, (pp. 295-324). Cambridge: Cambridge
University Press.
Eysenck, H. J. (in press). Personality and the biosocial model of antisocial and criminal
behavior. In A. Raine, P. Brennan, D. P. Farrington, & S. A. Mednick (Eds).
Biosocial bases of violence. New York: Plenum.
Eysenck, H. J (1964). Crime and personality (1st. ed.) London: Methuen.
Eysenck, H. J. (1977). Crime and personality (3rd. ed.). St. Albans, U. K.: Paladin.
Eysenck, H. J. (1987). Personality theory and problems of criminality. In B. J. McGurk,
David M. Thornton, & M. Williams (Eds.). Applying psychology to imprisonment-
Theory and practice (pp. 29-58). London: Her Majesty's Stationery Office.
Eysenck, H. J., & Eysenck, S. G. B. (1978). Psychopathy, personality and genetics. In R.
D. Hare, & D. Schalling (Eds.), Psychopathic behavior: Approaches to research (pp.
197-224). New York: Wiley.
Eysenck, H. J., & Gudjonsson, G. H. (1989). The causes and cures of criminality. New
York: Plenum.
Farrington, D. P., Biron, L., & LeBlanc, M. (1982). Personality and delinquency in
London and Montreal. In J. Gunn, & D. P. Farrington (Eds.), Abnormal offenders,
delinquency, and the criminal justice system. Chichester, U.K.: Wiley.
Hare, R. D. (1978). Electrodermal and cardiovascular correlates of psychopathy. In R.
D. Hare, & D. Schalling (Eds.), Psychopathic behavior: Approaches to research (pp.
107-144). New York: Wiley.
Hare, R. D. (1982). Psychopathy and physiological activity during anticipation of an
aversive stimulus in a distraction paradigm. Psychophysiology, 19, 266-271.
Hazlett, E., Dawson, M., Buchsbaum, M. S., & Nuechterlein, K. (1993). Reduced
regional brain glucose metabolism assessed by PET in electrodermal nonresponder
schizophrenics: A pilot study. Journal of Abnormal Psychology, 102, 39-46.
Hemming, J. H. (1981). Electrodermal indices in a selected prison sample and students.
Personality and Individual Differences, 2, 37—46.
140 Personality

Hugdahl, K., Berardi, A., Thompson, W. L., Kosslyn, S. M., Macy, R., Baker, D. P.,
Alpert, N. M., & LeDoux, J. E. (1995). Brain mechanisms in human classical
conditioning: a PET blood flow study. NewvReport, 6, 1723-1728.
Lykken, D. T. (1955). A study of anxiety in the sociopathic personality. (Doctoral
Dissertation, University of Minnesota), Ann Arbor, MI: University Microfilms, 1955,
No. 55-944.
Mednick, S. A. (1977). A bio-social theory of the learning of law-abiding behavior. In S.
A. Mednick, & K. O. Christiansen (Eds.). Biosocial bases of criminal behavior. New
York: Gardner Press.
Moffitt, T. E. (1993). Adolescent-limited and life-course persistent antisocial behavior:
A developmental taxonomy. Psychological Review, 100, 674-701.
Passingham, R. E. (1972). Crime and personality: A review of Eysenck's theory. In V.
D. Nebylitsyn, & J. A. Gray (Eds.). Biological bases of individual behavior. New York:
Academic Press.
Quay, H. C, & Parsons, L. B. (1970). The differential classification of the juvenile
offender. Washington, DC: Bureau of Prisons
Raine, A. (1993) The psychopathology of crime: Criminal behavior as a clinical disorder.
San Diego, CA: Academic Press.
Raine, A., Buchsbaum, M. S., Stanley, J., Lottenberg, S., Abel, L., & Stoddard, J.
(1994). Selective reductions in pre-frontal glucose metabolism in murderers.
Biological Psychiatry, 36, 365-373. (Also reprinted in J. A. Talbott (Ed.)(1996). The
Yearbook of Psychiatry and Applied Mental Health. St. Louis, MO: Mosby. (See also
correspondence in Biological Psychiatry, 1995, 38, 342-343).
Raine, A., Reynolds, G., & Sheard, C. (1991). Neuroanatomical mediators of
electrodermal activity in normal human subjects: A magnetic resonance imaging
study. Psychophysiology, 28, 548-558.
Raine, A., Venables, P. H., & Williams, M. (1995). High autonomic arousal and
electrodermal orienting at age 15 years as protective factors against criminal behavior
at age 29 years. American Journal of Psychiatry, 152, 1595-1600. (findings also
reported and discussed in Science, 1995, 270, 1123-1125).
Raine, A., Venables, P. H., & Williams, M. (1996). Better autonomic conditioning and
faster electrodermal half-recovery time at age 15 years as possible protective factors
against crime at age 29 years. Developmental Psychology, 32, 624-630.
Raine, A., & Venables, P. H. (1981). Classical conditioning and socialization—a
biosocial interaction? Personality and Individual Differences, 2, 273-283.
Raine, A., Venables, P. H., & Williams, M. (1990). Relationships between CNS and
ANS measures of arousal at age 15 and criminality at age 24. Archives of General
Psychiatry, 47, 1003-1007.
Scarpa, A, Raine, A., Venables, P. H, & Mednick, S. A. (1995). The stability of
inhibited/uninhibited temperament from ages 3 to 11 years in Mauritian children.
Journal of Abnormal Child Psychology, 23, 607-618.
Siddle, D. A. T. & Traster, G. (1981). The psychophysiology of psychopathic behavior.
In M. J. Christie and P. G. Mellett (Eds.). Foundations of psychosomatics (pp. 283-
303). Chichester: Wiley.
Stoddard, J., Raine, A., Bihrle, S. & Buchsbaum, M. (in press). Prefrontal dysfunction
in murders lacking psychosocial deficits. In A. Raine, P. Brennan, D. P. Farrington,
& S. A. Mednick (Eds), Biosocial bases of violence. New York: Plenum.
 Classical conditioning, arousal, and crime: A biosocial perspective 141

Tharp, V. K., Maltzman, I., Syndulko, K., & Ziskind, E. (1980). Autonomic activity
during anticipation of an aversive tone in non-institutionalized sociopaths.
Psychophysiology, 17, 123-128.
Venables, P. H. (1978). Psychophysiology and psychometrics. Psychophysiology, 15, BO-
SIS.
Wadsworth, M. E. J. (1976). Delinquency, pulse rate and early emotional deprivation.
British Journal of Criminology, 16, 245-256.
Ziskind, E., Syndulko, K., & Maltzman, I. (1978). Aversive conditioning in the
sociopath. Pavlovian Journal of Biological Science, 13, 199-205.
Chapter 8
Crime and personality

G. H. Gudjonsson

1. INTRODUCTION

In this chapter the author discusses Eysenck's theory of criminality, its

development chronologically over the past 35 years, some of the research that
has resulted from the theory, the empirical evidence for the theory, the theory's
strengths and weaknesses, and possible future developments. The focus will be
on evaluating the contributions Eysenck's theory has made to understanding
the causes and cures of criminality.

2. EYSENCK'S THEORY OF CRIMINALITY

Eysenck's theory rests heavily on the assumption that personality traits are
important predispositions to criminality. In its current form, the three major
and independent dimensions of personality—extraversion (E), neuroticism (N)
and psychoticism (P)—are considered by Eysenck to be the most fundamental
personality factors which predispose people to criminal conduct. E, N and P all
have, according to Eysenck, a strong hereditary and biological basis. E and P
are linked to criminality through low cortical arousal, poor conditionability,
and the failure to develop the conditioned response conscience. N, in contrast,
is associated with emotional instability and strong autonomic arousal, which
can lead to impulsive and antisocial behavior by virtue of its drive propensities.
Eysenck's primary theoretical goal is the prediction and control of criminal
behavior. In spite of Eysenck's emphasis on personality and biological factors,
he does not ignore the social causes of crime, but considers current sociological
theories to be scientifically untestable, and even if they could be proven to
exist, they must act through psychological pathways (Eysenck, 1996a).
Eysenck fully recognizes that criminality is a social concept, not a biological
one (Eysenck, 1977; Eysenck & Gudjonsson, 1989). Indeed, human behavior,
including the propensity to criminal behavior, is seen as being powerfully
influenced by both heredity and environment (Eysenck, 1964b). What
 Crime and personality 143

Eysenck's theory attempts to do is to explain why in a similar situation and

circumstance one person commits a crime while another does not. The main
hypothesis postulated by Eysenck's theory is that children, adolescents, and
adults who exhibit antisocial or criminal behavior will score higher than their
normal counterparts on tests measuring the traits of psychoticism, extra-
version, and neuroticism.

3. THE BASIS AND CHRONOLOGY OF EYSENCK'S THEORY OF CRIMINALITY

The earliest work of Eysenck into the relationship between crime and
personality dates back to his books The Scientific Basis of Personality (Eysenck,
1952) and The Psychology of Politics (Eysenck, 1954), where he puts forward
the hypothesis that extraverts are "prone to such disorders as hysteria and
psychopathy, i.e. asocial illnesses in which moral rules tend to be disregarded"
(Eysenck, 1954, p. 175). In contrast, introverts were liable to develop such
disorders as anxiety, depression, and obsessional-compulsive disorders. The
basis for Eysenck's arguments was modern learning theory, which postulated
fundamental individual differences in conditionability. These individual
differences, which had been documented by Pavlov in his landmark
experiments with dogs, were in humans linked to an important higher order
personality trait, namely introversion-extraversion. At the time, Eysenck
linked extraversion to toughmindedness and introversion to tenderminded-
ness. Toughmindedness was found to be associated with aggression and
dominance. Toughmindedness was later to feature in the concept of "Psycho-
ticism" or "P" as a personality trait which proved to be independent of the
extraversion and neuroticism dimensions as measured by the Eysenck
Personality Questionnaire (H. J. Eysenck & S. B. G. Eysenck, 1970, 1975).
Eysenck's initial ideas on crime and personality were developed further in a
commissioned article on the moral values in children for the British Journal of
Educational Psychology (Eysenck, 1960). The main hypothesis put forward in
the article was that "differences in conditionability determine in part the
socialized or anti-social behavior of children and adults alike" (p. 17). Rather
than asking the traditional question "Why do people commit crimes?" Eysenck
cogently argued that we should focus on "Why and how do human beings learn
to act in conformity with the dictates of society?" (p. 12). In other words, what
is it that prevents people from offending? He stated that fear or threat of
punishment is unlikely to be effective in deterring antisocial behavior due the
fact that punishment often occurs long after the event, if at all, whereas the
reward of the antisocial act was immediate and certain. Learning theory
postulated "that the effectiveness of reward and punishment is an inverse
function of the time interval between the act and reinforcement, and a direct
function of the proportion of reinforcements" (p. 12). On this basis Eysenck
144 Personality

argued that the fear or threat of punishment was not the main reason for moral
behavior. He suggested a possible alternative, namely, that "conscience is a
conditioned anxiety response to certain types of situations and actions" (p. 13).
According to this theory, the development of moral or prosocial behavior was
produced because undesirable behavior was punished by significant others
(e.g., parents, teachers). The punishment implemented produces emotional
pain, fear, and autonomic disturbances, which become associated, by means of
Pavlovian or classical conditioning, with certain situations and actions. A
conditioned emotional reaction is then experienced whenever the person
encounters a similar situation and action in the future. This emotional
response conscience, is according to Eysenck, precisely what prevents people
from offending.
Eysenck's book Crime and Personality was originally published in 1964
(Eysenck, 1964a) and was revised in 1970 and 1977 (Eysenck, 1970, 1977). In
the revised editions Eysenck amplified his theory of criminality and provided
up-to-date empirical support for his theory. Accompanying the publication of
Eysenck's book in 1964 were a number of journal articles, which focused on the
biological basis of criminality (Eysenck, 1964b, 1964c, 1964d). Eysenck post-
ulated that conditionability was the foundation for the development of a
conscience and it was seen as independent of intelligence. Intelligence as a
possible predisposition to criminal behavior does not feature in Eysenck's early
work on criminality. The emphasis is very much on personality as a pre-
disposition to criminal behavior.
In Eysenck's original theory it was not specified what types of offense or
offender it specifically applied to. The theory seemed to apply to all antisocial
and criminal behavior. This criticism of the theory was addressed by
Passingham (1967, 1972) in his detailed critique of Eysenck's theory. In the
1977 revision of his book, Eysenck addresses this issue and states:
In general terms, we would expect persons with strong antisocial inclinations to have
high P, high E and high N scores; a similar expectation would be reasonable with
respect to criminals, although in a somewhat modified fashion. Some criminals, for
instance, may not be antisocial, in the common meaning of that term. Political
prisoners are an obvious case in point. Homosexuals are another; homosexuality was
a crime in England until recently, but is so no longer. Murderers (of the type that
used to be predominant until recently, i.e. done in the family, not the terror murders
and those associated with armed robbery, now so common) tend to be introverted
and repressed, until they suddenly break out of their shell. In addition, there are
certain quite large groups of criminals which are quite likely to be introverted rather
than extraverted. For instance, there is a large group of people characterized by
inadequate personality, rather dull and helpless who drift into crime not because they
are in any sense antisocial, but they simply cannot cope with the complexities of
modern life ... At the other end of the continuum we have the actively antisocial,
psychopathic criminal who almost glories in his criminal activities and seems bereft of
conscience or guilt feelings. (Eysenck, 1977, pp. 5&-S9)
 Crime and personality 145

In 1989 Eysenck joined forces with the present author to produce The
Causes and Cures of Criminality (Eysenck & Gudjonsson, 1989). Part One of
the book, which focused on the causes of crime, was written by Eysenck and
Part Two, which focused on prevention and treatment, was written by
Gudjonsson. In Eysenck's 1964 edition of Crime and Personality not much was
written about the effectiveness of psychological treatment with criminals. This
was partly corrected in the 1977 edition of his book. In the Causes and Cures of
Criminality three long chapters were devoted to the area of punishment and
treatment effectiveness. This review highlighted the fact that most research in
the area was data driven and lacked a sound theoretical basis. It was striking
that there was virtually no association between Eysenck's theory of criminality
and the development of treatment programs. Eysenck and Gudjonsson
presented a number of studies which indicated that intelligence was negatively
associated with antisocial and criminal behavior. One explanation put forward,
which was based on the work of Kendel et al. (1988) in Denmark, is that
intelligence can act as a protective factor for men who are at high risk of
becoming actively involved in criminal activity.
Eysenck's most recent publications on criminality (Eysenck, 1996a, 1996b)
focus on personality and a biosocial model of antisocial and criminal behavior
(see also chapter 6 in this volume). He concludes that violence and crime have
biological and social roots, including genetic factors which are linked with
behavior via hormones, physiological factors, and neurotransmitters. All of
these are, according to Eysenck, determinants of personality. Eysenck then
goes on to examine the relationship between biological factors, criminal
behavior, and personality. He does this by discussing the chain that links DNA,
personality, and criminality. The link starts with DNA, the basic genetic
structure which underlies individual differences. DNA, referred to as "distal
antecedents," influences behavior through biological intermediaries in the
central and autonomic nervous system, which are labeled "proximal ante-
cedents." These biological intermediaries or determinants are the immediate
antecedents for individual differences in personality, such as the traits of P, E,
and N. These higher order personality traits are made up of a number of
elementary traits and influence conditionability, sensitivity, vigilance, memory,
and reminiscence (referred to as "proximal consequences"). These in turn
interact with social conditions to produce a range of social behaviors, including
criminality, creativity, psychopathology, and sexual behavior. Eysenck's des-
criptions of the elementary traits that make up P, E, and N are as follows
(Eysenck, 1996a; Eysenck & Gudjonsson, 1989):
P (aggressive, cold, egocentric, impersonal, impulsive, antisocial, unempathic,
creative, tough-minded);
E (Sociable, lively, active, assertive, sensation-seeking, carefree, dominant, surgent,
venturesome);
146 Personality

N (anxious, depressed, guilt feeling, low self-esteem, tense, irrational, shy, moody,
emotional).
P, which is Eysenck's most recent personality dimension, did not develop out
of theory of personality in the way that E and N did. Rather the dimension was
originally associated with criminality, because relatives of psychotic patients
were sometimes observed as having antisocial behavioral problems and
criminals were found to score high on the original P scale (H. J. Eysenck &
S. B. G. Eysenck, 1970, 1975).
An interesting aspect of Eysenck's most recent paper (Eysenck, 1996b) is his
apparently greater acceptance of the social causes of crime. In his early work
Eysenck never totally ignored the social causes of crime. However, he has
always been skeptical of sociological theories of crime, particularly the
economic theories (Eysenck, 1996b; Eysenck & Gudjonsson, 1989). In spite of
his skepticism, Eysenck recognizes that huge differences between countries in
the crime rate, and sudden changes within a given community like those in
Russia following the overthrow of communism, cannot be explained in terms of
genetic differences and personality factors. Here the nature of the social
constitution in the country concerned remains the most important explanatory
factor. Nevertheless, the social causes do not act in isolation to psychological
factors; they influence people's minds and produce psychological conditions
which are favorable to antisocial conduct (Eysenck, 1996b). The precise nature
of these psychological conditions and the salient mediating variables are not
examined by Eysenck, but in his previous work (Eysenck, 1977) he broaches
this briefly in relation to domestic killings and "inadequate personalities." This
is an area that requires a greater theoretical focus and empirical research.
What precisely are the psychological conditions which turn conscientious and
law-abiding citizens into criminals?

4. THE DEVELOPMENT OF A CONSCIENCE

The original link proposed by Eysenck between extraversion and criminality

was in terms of Pavlovian conditioning. The essence of Eysenck's theory is that
conditioning is aided by high cortical arousal. High arousal leads to good
conditioning and the development of a conscience. Conscience was viewed as
a conditioned response by means of positive and negative reinforcements of
prosocial and antisocial behavior, respectively. Low arousal results in poor
Pavlovian conditioning, followed by a lack of conscience and increased
likelihood of antisocial behavior.
Eysenck (1964a) originally argued that prosocial behavior was a function of
two important factors. First, the degree of conditionability of the person.
Second, the quality of conditioning applied during the person's childhood by
significant others, such as parents, teachers, and peers. Since introverts
 Crime and personality 147

condition more readily than extraverts, they are theoretically less predisposed
to antisocial and criminal behavior than extraverts, provided the quality of the
conditioning is satisfactory for the development of a conscience and prosocial
behavior.
In relation to the development of a conscience, Eysenck (1996a) recently
expands upon his original ideas and now discusses three possibilities, which are
not mutually exclusive. Eysenck's first point is that problems arise because the
conditioning experiences necessary for the development of a conscience are
missing. Here significant people in the child's life fail to punish undesirable
behavior and as a consequence no proper reinforcers are developed. Second,
wrong and undesirable experiences may be reinforced by significant others.
For example, parents may actively encourage their children to act aggressively
and dishonestly. Third, low cortical arousal makes conditioning less likely to
occur. P is now seen as having an important part to play in the development of
conscience, whereas in Eysenck's early work only E was mentioned. Eysenck
quotes the recent review of Raine (1993) on conditioning as evidence for the
poorer conditionability of criminals and people with antisocial tendencies.
Raine and Venables (1981) rightly argued that if Eysenck's theory was
correct then the outcome of the conditioning process would depend on the
quality of the conduct to which the person is being conditioned as well as on
the conditionability of the person. This is an important point, because readily
conditioned persons, such as introverts, may develop a conditioned response to
undesirable behaviors. For example, a child who is sexually abused often
becomes an abuser of children when he or she grows up (Salter, 1988). A
similar pattern is seen in those children who are physically abused. The
modeling of aggressive, and also possibly sexual, behavior is probably an
important factor in these cases (Patterson, 1986). The point is that perversions,
whether sexual, physical, or emotional, can be subjected to a similar con-
ditioning process as prosocial behavior. Applying Eysenck's theory, in per-
versive social environments introverts may be more adversely affected than
extraverts to unadaptive conditioned responses due to their greater condition-
ability. This may explain why paedophiles tend to be introverted (Wilson &
Cox, 1983). Similarly, serial murderers typically come from dysfunctional
families which involved sexual, physical, or emotional abuse (Douglas &
Olshaker, 1996). Although no empirical study has been conducted into the
personality of serial murderers, many of them appear to be introverted
(Holmes & De Burger, 1988; Sears, 1991).
From his earliest work on personality and crime, Eysenck has placed much
importance on the role of conditioning in the development of conscience. The
work of Solomon and his colleagues (Solomon, Turner, & Lessac, 1968) was
influential in amplifying Eysenck's theory. On the basis of their work with dogs,
Solomon and his colleagues assumed that conscience was made up of two
components, one was resistance to temptation and the other was the
148 Personality

susceptibility to guilt reactions after transgression. These represent two distinct

forms of socialization, which are linked to the timing when punishment is
implemented for transgression. Resistance to temptation in the child is
developed when the parent punishes the child immediately prior to the
transgression (i.e. during the incipient stage). This involves the parent watching
the child's behavior closely and being able to identify intent (i.e. clearly
identifying what the child is intending to do). This has important deterrent
effects in terms of behavioral control. In contrast, when the punishment occurs
after the transgression then this produces emotional reactions of guilt.

5. IMPLICATIONS FOR PREVENTION AND TREATMENT

In his early work, Eysenck (1964a, 1964b) discussed the two types of approach
traditionally used to deal with criminal behavior, which related to using
punitive or psychotherapeutic measures, respectively. The psychotherapeutic
measures, Eysenck argued, were based on unproven theories and lacked
empirical support, and greater severity of punishment was both cruel and
inefficient. He states:
At the bottom of all our errors lies probably a fundamental psychological fallacy. We
think that punishment deters, and we go on to imagine that the more severe the
punishment the greater the deterrence. This is not always true. The statistical study of
the effects of increasing the severity of punishment, as by corporal punishment, death
penalty, etc. is in complete accord with experimental laboratory studies in showing
that the effects of punishment are extremely variable, very difficult to predict and
often contradictory to expectation. Severe punishment heightens the prisoners'
emotionality to a very considerable degree and this 'booster' action may combine with
the existing systems of habits to make these more rigid and difficult to eliminate.
These facts have been demonstrated so often as to constitute a psychological truism,
yet the only action most people can think of when confronted with an increase in
crime is to call for greater severity of punishment, i.e. precisely the remedy least likely
to succeed and most likely to make matters worse. (Eysenck, 1964b, p. 10)
The above quote indicates that over 30 years ago Eysenck was strongly
opposed to punishment as a way of preventing crime. Nor did he have much
faith in the effectiveness of psychotherapeutic techniques applied to crime
prevention. What was Eysenck s solution to the crime problem when he
entered the field of criminology in the early 1960s? His most basic and
fundamental assumption was that individual differences in personality imply
that matching the treatment with the personality of the offender will produce
the best outcome results in terms of crime prevention. The next step in
Eysenck's line of reasoning was that introverted and extraverted criminals
require different types of treatment. He maintained that extraverts, due to
their innate deficit in conditionability, will be responsible for most criminal
 Crime and personality 149

acts. One way of dealing with the inherent criminal potential of extraverts is by
changing their position on the extravert-introvert continuum. Eysenck argued,
in all three editions of his book Crime and Personality, that this could be done by
means of prescribed drugs. Stimulant drugs, such as nicotine, caffeine, and
amphetamine were said to have an introverting effect, whereas depressant drugs,
such as alcohol and barbiturates, have an extraverting effect. Eysenck suggested
that in order to facilitate the conditioning process when dealing with highly
extraverted persons, "we simply have to shift them along the continuum by daily
doses of a stimulant drug such as amphetamine" (Eysenck, 1964b, p. 10).
To support his argument, Eysenck (1964a) quotes four studies where
stimulant drugs were shown to improve behavior. In the first study, Lindsley
and Henry (1942) showed the effects of the administration of a stimulant drug
(benzedrine) and a depressant drug (phenobarbital) on the behavior of 13
children with behavioral disorder. The benzedrine improved behavior, whereas
the phenobarbital exacerbated their antisocial behavior. Eysenck quotes a
study by Cutts and Jasper (1939), who studied 12 children with behavioral
problems. Benzedrine markedly improved the behavior of half the children,
whereas when the children were given phenobarbital their behavior
deteriorated in nine (75%) of the cases. These findings are similar to those
found by Lindsley and Henry. In another study, Bradley and Bowen (1941)
investigated the effects of the stimulant drug amphetamine on the behavior of
100 children. In 54% of cases the children became more "subdued," although
some of the other children became more active and aggressive. In a group of
normal subjects, Franks and Trouton (1958) showed that the stimulant drug
dexadrine facilitated conditioning in an eye-blink experiment, whereas the
depressant drug sodium amytal depressed conditioning.
None of the experiments with the behavioral disordered children combined
the drug administration with a planned program of conditioning, which
according to Eysenck, would probably have improved the behavioral outcome
further. Eysenck's suggestion about the use of stimulants in the treatment of
behavioral disorders to increase their introversion while implementing a
systematic program of conditioning has not met with much enthusiasm.
Indeed, the present author is not aware of any published outcome studies in
this area. It is unlikely that Eysenck's controversial recommendation would
meet the approval of current ethical committees. However, Eysenck does have
a point in focusing our efforts for crime prevention on children with behavioral
disorders. Children who exhibit symptoms of hyperactivity, impulsivity, and
attention problems are particularly at risk of becoming repeated offenders in
adulthood and this is the group of children who we should treat before they
reach adulthood (Lynam, 1996).
In the 1977 edition of Crime and Personality Eysenck expanded upon his
ideas of treatment interventions and introduced the concept and measurement
of psychoticism (P). The treatment recommendations involved the application
150 Personality

of modem learning theory to two aspects of the criminal's behavior. First, the
elimination of undesirable conditioned responses and second, the creation of
new more desirable conditioned responses. With regard to the former,
Eysenck states:
Other problems, of course, require other methods and no one would argue that the
success of this particular treatment, which has been duplicated with many other
similar cases as well as with transvestites, homosexuals and others, can necessarily be
expected with the typical kinds of criminals and delinquents we have in our prisons.
What it demonstrates, however, is that by a suitable experimental conditioning
regime we can decondition a very powerful impulse to perform a certain act, in this
particular case, an aggressive and illegal one; and that, therefore the theory on which
the treatment is based holds considerable promise, even in regard to other types of
criminal conduct, to which it has not yet been applied. (Eysenck, 1977, p. 170)
The case referred to by Eysenck involved a man with a history of damaging
prams and handbags for sexual gratification. The aim of treatment was "to
alter his attitude to handbags and perambulators by teaching him to associate
them with an unpleasant sensation instead of with a pleasurable, erotic
sensation" (p. 169). The treatment involved injecting the patient with apo-
morphine and inducing nausea just before he was shown handbags and prams.
The treatment was successful in eliminating the undesirable behavior and at
the same time improved his sex life with his wife.
Eysenck emphasized the need for creating desirable conditioned responses
for improved prosocial behavior rather than focusing predominantly on the
elimination of undesirable conditioned responses. He argued that one good
example from the application of modern learning theory was the successful
treatment of enuresis in children by the use of the "bell and blanket" method
of Mowrer. Eysenck states:
Again, it is not suggested that this method can be applied directly to the kind of
problems presented by the criminal. It is cited here merely to indicate that, with
sufficient ingenuity, it is possible to make deductions from learning theory, which
may lead to a solution of these problems. What may be the most appropriate method
for the treatment of criminals remains to be seen. (Eysenck, 1977, p. 171)
Eysenck then goes on to provide some suggestions for the future and argues
that aversion therapy is unlikely to be effective with some offenders
because the stimuli which would have to be deconditioned would be so varied as to
make the task impossible. One can use aversion therapy in connection with very
specific stimuli (drink, prams, or even homosexuality), but not with such a broad and
nebulous concept as criminality. A much more appealing alternative is rather to
strengthen those positive aspects of behavior which we have provisionally grouped
under the concept of "conscience." The method for achieving such an ambitious aim
which learning theory suggests has been given the name "token economy," and there
are already some positive achievements to its credit. (Eysenck, 1977, p. 175)
 Crime and personality 151

6. THE EVIDENCE FOR EYSENCK'S THEORY

A number of critical reviews of Eysenck's theory have been put forward

(Blackburn, 1993; Gray, 1981; Passingham, 1967, 1972; Trasler, 1978). Other
writers have focused their review specifically on Eysenck's third dimension—
psychoticism (Powell, 1977; Howarth, 1986; Zuckerman, 1989), or upon the
empirical evidence (Cochrane, 1974; Fonseca & Yule, 1995; Gabrys et al., 1988;
Lane, 1987; Raine, Venables, & Williams, 1996; Rushton & Christjohn, 1981).
Passingham, an ex-student of Eysenck's, completed his M.Sc. dissertation on
Eysenck's theory in 1967. In an early review (Passingham, 1972), he cogently
and comprehensively reviewed the merit of the theory and reviewed the
relevant empirical evidence. Passingham was highly critical of some aspects of
Eysenck's theory. His main criticisms are as follows:
1. Eysenck does not specify what types of offense his theory applies to. Is it equally
applicable to all offenses, including violent, sexual, property, drug, and traffic
offenses?
2. It is not clear if Eysenck's theory is applicable to all types of offender, including
males, females, juveniles, first offenders, recidivists, psychopaths, etc.
3. There are different types of learning involved in producing criminal behavior,
including classical conditioning, instrumental (operant) conditioning, imitative
learning (the role of modeling), and human verbal learning (there is also "passive
avoidance learning," which is not mentioned by Passingham but will be discussed
below). Passingham argues that there is lack of evidence for a general factor of
conditionability across different types of learning. Eysenck's theory relies on a narrow
concept of human development which is derived from studies conducted in animal
laboratories.
4. Eysenck does not provide evidence that introverts are better than extraverts in
learning positive instrumental responses and focuses almost exclusively on classical
conditioning. However, Eysenck recognizes the importance of both classical and
operant conditioning and states that they "are probably rather different principles of
learning, although it is often not easy to know which principle is active" (Eysenck,
1977, p. 132).
5. Passingham argues that two of the conditions which Eysenck says favor
conditioning in introverts, weak unconditioned stimulus and short intervals between
the conditioned and unconditioned stimulus, are not common in everyday life,
whereas the other two conditions, partial reinforcement and discrimination learning,
are probably fulfilled.
6. The empirical data on the personality and conditionability of delinquents and
criminals is unsatisfactory due to methodological problems with many of the studies.
There are also examples of different studies providing contradictory conclusions,
possibly due to differences in the samples or controls used. "Even when the
inadequacy of the studies is taken into account, this review indicates that the evidence
on Eysenck's theory is only partially confirmatory, suggesting that the theory applies
only, if at all, to a subgroup of offenders. This subgroup may be related to the
subgroup with the syndrome of aggressive psychopathy" (Passingham, 1972, p. 365).
152 Personality

In spite of Passingham's criticisms of Eysenck's theory, he clearly recognized

Eysenck's unique contribution to the study of individual differences. He
pointed out that Eysenck's book Crime and Personality presented researchers
with a theory and testable hypotheses which could guide future research. Prior
to Eysenck's theory, previous research into crime and personality lacked a
theoretical focus and testable hypotheses (see Schuessler & Cressey, 1950, for
a review of previous research in the field). Schuessler and Cressey had
reviewed the use of 30 personality tests in differentiating between criminal
(prison inmates) and noncriminal groups over a 25-year period. Research into
a "criminal personality" had been growing in popularity at the time, but the
results from the studies reviewed were disappointing. The authors concluded
that personality tests were not useful measures in criminological research.
In his review of Eysenck's theory, Blackburn (1993) argues that the most
powerful explanatory component of the theory is the linking of E with low
cortical arousal and lack of socialization. In contrast, he argues, N and P
possess little explanatory power. P is not derived from theory, but from the
empirical finding that criminal behavior and psychopathy is more common
among the relatives of psychotic patients. Blackburn argues that if the
impulsivity part of E is of theoretical significance in Eysenck's model, but has
now been largely transferred to P, then this seriously undermines his theory of
criminality. It appears from Eysenck's (1996b) most recent work that he now
links P with low cortical arousal, poor conditionability, and lack of
socialization, along similar lines he originally detailed for E. The crucial
trait, in Eysenck's view, linking conditioning to personality is impulsivity. This
does not mean that extraversion, in its revised measured form and focusing less
directly on impulsivity, is no longer of relevance to conditionability and
criminality. It still remains a feature of E, although perhaps of a different type.
Eysenck has clearly redefined his theory of criminality to accommodate the
increased recognition that high P appears to be an important hindrance to
effective socialization. This illustrates the close relationship between theory
construction and questionnaire development, which has been so influential in
Eysenck's work on impulsivity (Gray, 1981).
Gray (1981) presents some criticisms of Eysenck's theory and demonstrates
subtle changes in the theory over time. His criticism focuses largely around the
E dimension and the arousal theory put forward to explain it. The first
problem, according to Gray, is the inability of Eysenck's theory to explain time-
of-day effects on E, which Gray suggested was due to the impulsivity
component of E. The second criticism is that findings of superior performance
of introverts in eye-blink conditioning does not mean that we can generalize to
the types of conditioned responses that make up conscience. Third, as far as
socialization is concerned, Gray sees conditioning as being important to the
extent that introverts are biologically more susceptible to fear reactions than
extraverts and condition more readily with an aversive unconditioned stimulus.
 Crime and personality 153

Fourth, the impulsivity part of E correlated better with conditioning than the
sociability component. Removing impulsivity to P, with possibly different types
of impulsivity does not, according the Gray, solve the problem with Eysenck's
theory, because the new-style impulsivity of P no longer has the biological
correlation with diurnal rhythms as did the old-style impulsivity.
Gray goes on to postulate his own theory to overcome the problems with
Eysenck's theory. He rotates Eysenck's E and N dimensions by 45 to produce
two new dimensions, labeled "anxiety" and "impulsivity," which are related to
differences in susceptibility to reward and punishment, respectively. According
to Gray's theoretical framework, introverted neurotics (high N, low E) are
highly susceptible to punishment, whereas neurotic extraverts (high N, high E)
are most susceptible to reward. In relation to conditioning, introverts only
condition better than extraverts if their introversion is accompanied by high
arousal (N), which amplifies the strength of the aversive event and increases
the introvert's susceptibility to fear.
Impulsivity and anxiety are viewed by Gray as representing two funda-
mentally different systems of learning. The dimension of impulsivity is seen as
reflecting individual differences in the activity of the "behavioral approach
system" (BAS), whereas anxiety relates to activity in the "behavioral inhibition
system" (BIS).
The implication of Gray's theory is that antisocial persons would in general
be more responsive to reward and less sensitive to punishment than normal
controls. There is some empirical support for this hypothesis (Fonseca & Yule,
1995). Eysenck and Gudjonsson (1989) argue that Gray's theory has important
implications for psychological treatment programs. Neurotic extraverts are
unlikely to respond greatly to aversive-conditioning procedures aimed at
reducing undesirable behaviors, whereas they would be more responsive to the
reinforcing of alternative (prosocial) behavior. Therefore, Gray's theory may
guide us as to whether the focus of treatment of criminal behavior should be
directly on the elimination of undesirable conditioned responses, or
alternatively focus more on the creation of new more desirable conditioned
responses. Gudjonsson (1987) presents a treatment case of a compulsive
shoplifter which supports this important distinction.
Zuckerman (1989) argues that P is a much more complicated dimension
than E and N and includes such traits as impulsivity, lack of socialization,
aggression, strong need for independence, and sensation seeking. Due to its
association with lack of socialization and impulsivity, Zuckerman argues that P
is probably more directly associated with conditionability than E and has a
strong biological basis related to a low level of the neurotransmitter serotonin
and deficits in neuroregulating enzymes. Zuckerman sees lack of condition-
ability as being more directly related to a narrow impulsivity dimension rather
than to the broader aspects of the P, E, and N dimensions. He emphasizes the
importance of deficient passive avoidance learning in psychopaths as being
154 Personality

associated with their antisocial behavior. Passive avoidance learning, which is

distinct from both classical and instrumental conditioning, refers to cues
associated with punishment or loss of reward arousing conditioned anxiety
which is reduced when restraining action is exercised. In his early work on
psychopaths, Lykken (1957) demonstrated the psychopath's poor passive-
avoidance, which has been repeatedly supported by other studies (Lynam,
1996).
Passive avoidance learning forms an important part of Trasler's (1978,1979)
theory of socialization, but it does not feature at all in Eysenck's work. Like
Eysenck he sees conditioned anxiety as a more important aspect of sociali-
zation than the perceptions of aversive contingencies in the person's immediate
environment, but unlike Eysenck he relates the conditioned anxiety directly to
passive avoidance learning.
Passive avoidance learning has been used to explain why extraverts are less
responsive to punishment than introverts. According to Eysenck and
Gudjonsson (1989), extraverts are particularly unresponsive to punishment
when they are required to inhibit behavior which has previously been
associated with reward.

7. EMPIRICAL STUDIES

Many of the studies conducted into the relationship between crime and
personality over the past 30 years have tested hypotheses generated by
Eysenck's theory. The different forms of his personality questionnaire have
most commonly been used to test his theory. The results from these studies
have been equivocal. Some studies give full support for Eysenck's theory that
criminals have higher P, E, and N scores than nonoffenders (e.g., Eysenck,
1977; Gabrys et al., 1988), while other studies have found little if any support
for the theory (e.g., Farrington et al., 1982; Fonseca & Yule, 1995).
In order to overcome some of the inconsistent findings from the previous
research, McGurk and McDougall (1981) used cluster analysis to demonstrate
the heterogeneity of personality types in criminals and normal controls. Two
personality types were found in the offender group, the first being persons
scoring high on N and E, and the second type scoring high on P, E, and N. The
authors concluded that these findings gave support for Eysenck's theory and
point to the importance of subtypes of criminal personalities.
McEwan (1983) researched further the idea of different subgroups of
offenders by the use of cluster analysis. The results corroborated the findings
of McGurk and McDougall (1981) that there are different personality types
within criminal populations which are consistent with Eysenck's theory. The
most predictive type was that characterized by a high P scorer and it was
associated with the number of previous convictions. Type of offense was not
 Crime and personality 155

found to be a significant variable in this study. In contrast, Berman and Paisey

(1984) found that juveniles convicted of violent offenses scored significantly
higher on P, E, and N than those convicted of property offenses. These
findings suggest that it is the persistence and extent of offending which is most
highly related to Eysenck's theory of criminality. This conclusion is also
supported by research among delinquent children (Lane, 1987). Again, of the
three personality dimensions P is most discriminative.
S. B. G. Eysenck, Rust, and H. J. Eysenck (1977) studied the personality
traits of five different groups of criminals. Con men were most different from
the other groups, scoring low on P and N, but high on E, whereas criminals
described as "inadequate" on the basis of their criminal background scored
high on P and N, but rather low on E. Violent and property offenders were also
rather high on P and N.
The idea of personality heterogeneity of offenders and the use of cluster or
discriminant analysis for differentiating them into subgroups is an interesting
development, which could be used fruitfully in future research. This method
has recently been applied successfully to understanding the personality of high
risk takers (Goma-i-Freixanet, 1995) and patients who attempted suicide
(Engstrom, Alsen, Gustavsson, Schalling, & Traskman-Bendz, 1996).
In a major review of Eysenck's theory of criminality, Farrington et al. (1982)
concluded that of the three dimensions P is most commonly related with
delinquency and criminality as measured by either criminal convictions or self-
report measures. N often appears to be elevated in offender samples,
particularly among prisoners, but it is not known to what extent this is an
artifact of the effects of incarceration. E may also be influenced by the effects
of incarceration in that it may decrease the score due to restriction of
prisoners' being able to socialize in that setting. This raises some fundamental
questions about the validity of personality measures, such as the EPQ, when
applied to people whose liberty is restricted. The failure to find a significant
relationship between Eysenck's personality dimensions and criminality may
reflect inherent problems with the measures used, raise fundamental problems
with Eysenck's theory, or it involves a combination of both.
A recent prospective longitudinal study by Raine, Venables, and Williams
(1995,1996) demonstrates the importance of autonomic reactivity as a possible
biological protective factor against recidivism. A total of 101 subjects were
tested psychophysiologically when aged 15 years, and at age 29 years their
criminal records were obtained. Seventeen (17%) of the subjects had a
criminal conviction for a serious offense and had been delinquent at the age of
15 years. Their psychophysiological responses when aged 15 were compared
with 17 "desisters" from the original sample, who were delinquent adolescents
but had no adult criminal convictions. Another "control" group involved 17
subjects who had no adult criminal convictions and were not delinquent as
adolescents. Raine et al. (1995) found that the desisters had higher autonomic
156 Personality

(electrodermal and cardiovascular) arousal and electrodermal orienting

responses than the criminal control group. In a subsequent paper, Raine et
al. (1996) expanded their analyses to investigate if conditioning and electro-
dermal conductance recovery times (fast fear dissipation) also differentiated
between the groups, which they did in the predicted direction. The findings
give strong support for the authors' conclusion that biological (psychophysio-
logical) factors may be an important protection against adult criminal behavior
among delinquent adolescents. The precise protective mechanism is not fully
understood, but it may involve, as Eysenck's theory suggests, good condition-
ability facilitating the development of a conscience.
Intellectual factors have also been shown to have protective potential against
criminal behavior in high-risk groups (Kendel et al., 1988). Kendel et al.
suggest that the mechanism may involve the reinforcing effects of school
success in intellectually bright children. Of course, there are a number of
factors beside personality and intelligence which may act as early predictors of
adult criminal behavior, including family criminality, poor child-rearing, anti-
social childhood behavior, economic deprivation, and hyperactivity-impul-
sivity-attention deficit (Farrington, 1989). Various biological factors, such as
neurotransmitter abnormalities, are increasingly being recognized as being
important factors that predispose some persons to persistent criminality and
may explain very marked sex differences in crime rates (Moir & Jessel, 1995).
Many criminals are difficult to treat. High P scorers have been shown to be
particularly resistant to psychological treatment interventions (Lane, 1987;
Rachman & Eysenck, 1978). Rachman and Eysenck hypothesized that high P
scorers would be resistant to psychotherapy and behavior therapy because of
then- hostility, uncooperativeness, quarrelsomeness, and paranoia. The
patients were rated by their doctor for treatment outcome. The hypothesis
was supported among male and female patients treated for various neurotic
disorders. This finding was corroborated by a study of behavioral therapy
responsiveness in children (Lane, 1987). Lane found that P was negatively
correlated with both short- and long-term success in therapy. These two studies
suggest that a high P score predicts therapy outcome. P scorers may therefore
be hard to treat by the use of psychotherapeutic techniques.

8. ATTRIBUTION OF CRIMINAL BLAME AND PERSONALITY

In recent years there has been increased recognition that offenders' perception
of their offense hi terms of blame is related to personality. Gudjonsson (1984)
devised a "Blame Attribution Inventory" which measured two types of
attribution relevant to how offenders attribute blame for their criminal acts.
First, "external attribution," which measures the extent to which offenders
report external justification for their crime (e.g., blaming the offense on
 Crime and personality 157

provocation or social factors). Second, "mental element attribution," which

measures the extent to which offenders blame their crime on mental factors,
such as low mood and a temporary loss of self-control. The inventory also
measured the amount of guilt or remorse that offenders reported feeling about
the offense they had committed. The Blame Attribution Inventory was revised
1989, retaining the three factors, external attribution, mental element
attribution, and guilt feeling (Gudjonsson & Singh, 1989).
Three studies have been conducted into the relationship between person-
ality, as measured by the EPQ, and how offenders attribute blame for their
criminal acts according to the Blame Attribution Inventory. In the first study
involving 40 forensic patients in England, Gudjonsson (1984) found that
external attribution of blame correlated significantly with P, using the original
Blame Attribution Inventory. In another study of 169 English prisoners,
Gudjonsson and Singh (1989) found that external attribution correlated
significantly with P, using the revised version of the Blame Attribution
Inventory. Introversion and neuroticism were significantly associated with how
much remorse the criminal reported feeling about the offense. In the third
study, conducted on 68 prison inmates in Iceland, Gudjonsson, Petursson,
Sigurdardottir, and Skulason (1991) found that P correlated significantly with
external attribution on the revised Blame Attribution Inventory. Again,
remorse was significantly correlated with introversion and neuroticism. As far
as mental element attribution is concerned, there were positive correlations
with neuroticism in two of the studies (Gudjonsson, 1984; Gudjonsson et al.,
1991).
In the Gudjonsson (1984) and Gudjonsson et al. (1991) studies, the Gough
Socialization Scale (Gough, 1960) was also administered to the offenders, but
no significant relationship was found with external attribution of blame in
either study. This raises some important questions about the nature of P. It
suggests that it is the tough-minded and unempathetic criminal who has the
greatest tendency to blame other people or circumstances for his crime.
Emotional coldness and lack of empathy are probably the most relevant
characteristics to external attribution of blame. These are probably more
directly measured by P than the Gough Socialization Scale. This conclusion is
consistent with the work of Schalling (1978) into the validity of the Gough
Socialization Scale among Swedish criminals. The findings that remorse was
positively correlated with introversion and neuroticism and negatively with
psychoticism support Eysenck's theory of the development of conscience as a
conditioned response.
The work of Gudjonsson and Roberts (1983) into feelings of guilt and self-
concept in "secondary psychopaths," who were being treated in a therapeutic
community in England, raises some important questions about the nature of
guilt. Twenty-five male and 25 female psychopaths were compared with normal
subjects with regard to "Morality-Conscience Guilt" as measured by the
158 Personality

Mosher True-False Guilt Inventory (Mosher, 1966). Surprisingly, the

psychopaths scored significantly higher on the guilt inventory than the normal
subjects. The Semantic Differential technique (Osgood, Suci, & Tannenbaum,
1957) was used to measure the subjects' self-concept on 10 bipolar dimensions.
The subjects rated themselves on the 10 bipolar dimensions with regard to the
following concepts: "Myself as I am," "Myself as I would like to be," Myself
when I lie," "Myself if I were to steal," "People who lie," and "People who
steal." Factor analysis of the 10 bipolar dimensions revealed three factors,
referred to as an "evaluative" (good-bad), "potency" (strong-weak), and
"guilt" (remorseful-unremorseful, and ashamed-unashamed) factors, respect-
ively. The psychopaths had significantly greater semantic distance between
their self and ideal self than the normal subjects on all three factors, which
reflects their poorer self-evaluation. As far as the guilt factor was concerned,
the normal subjects had very little discrepancy between their self and ideal self,
whereas the psychopaths wished to feel less guilt and shame. A particularly
interesting finding was that the normal subjects reported a marked increase in
their feelings of guilt and shame when they were transgressing (i.e. lying or
stealing), whereas in the case of the psychopaths there was no change in their
degree of guilty feeling when transgressing.
The findings suggest that some psychopaths have poor self-concept which is
reflected in strong inner turmoil and negative preoccupations, which they label
as guilt, regardless of whether or not they are involved in antisocial behavior.
In contrast, normal subjects only experience feelings of guilt when they violate
some norms. If some psychopaths do experience strong feelings of guilt, which
are unrelated to specific situational transgression, then this may explain why
their feelings of guilt fail to inhibit antisocial behavior. That is, engaging in
antisocial acts does not make them feel any worse than they already feel.
Assuming that the guilt reported by some psychopaths represents genuine
feelings of guilt, rather than mislabeling, how does one interpret their
apparently high degree of guilt? One possible explanation may relate to these
psychopaths being punished indiscriminately in childhood for both prosocial
and antisocial behavior. Irrespective of the moral value of their behavior, they
are punished. They can never do anything right in the eyes of significant
persons in their lives. They consequently develop a conditioned response to
their own behavior which becomes generalized rather than being situation
specific to legitimate transgression. Placing this within the framework of
Solomon and his colleagues, the psychopaths in the Gudjonsson and Roberts
study never learned to resist temptation due to the timing of the punishment
implemented in childhood, which resulted in strong indiscriminate emotional
responses to perceived punishment. The psychopaths in the Gudjonsson and
Roberts study were labeled as "secondary psychopaths" in view of their high
level of trait anxiety and physiological reactivity (Gudjonsson & Roberts,
1985). Mealey (1995) has highlighted some important differences between
 Crime and personality 159

"primary" and "secondary psychopaths," the main difference being that latter
are primarily antisocial due to exposure to environmental risk factors, whereas
the antisocial behavior in primary psychopaths is primarily determined by their
genotype (i.e. have a substantial genetic component). Another difference
between the two types of psychopaths is that secondary psychopaths, in
contrast to primary psychopaths, are capable of experiencing some sincere
social emotions (e.g., guilt, shame, sympathy, empathy).
Eysenck (1995), in his peer review commentary on Mealy's target article,
criticizes the two categorical (typological) classes of psychopaths, primary and
secondary. He has long argued that categorical diagnosis is fundamentally
wrong and should be replaced by a dimensional system. Eysenck points to the
low reliability commonly associated with categorical diagnoses as a major
problem and argues for a more natural and reliable dimensional system. His
own preference is to view both primary and secondary psychopaths along a
three-dimensional framework, represented by elevations on E, N, and P.
Primary psychopaths would be particularly high on P.

9. CONCLUSIONS

Eysenck's scientific contribution to psychology, including criminological

psychology, is immense and unique. He introduced a scientific and objective
approach to the study of personality and individual differences. Prior to
Eysenck's book Crime and Personality in 1964, there was no clear theoretical
base from which research in the area could be developed. That was all to
change with Eysenck's landmark book.
Eysenck's starting point was that from psychological laboratories, theories
and experimental procedures had emerged which could be applied to criminal
behavior. Of considerable influence was Pavlov's work which had demon-
strated experimentally that there were individual differences in condition-
ability. Eysenck's earliest work on crime and personality was to relate
individual differences in conditionability to the development of the condition-
ed reaction conscience, which was seen as the basis for prosocial behavior.
Introversion facilitated the conditioning process, whereas extraversion (E)
slowed it down. Another important influence on Eysenck's theory was Hull's
theory that drive interacts with habit strength to augment existing responses.
Therefore, emotionally labile persons (high N) with antisocial tendencies will
engage in these behaviors more strongly than emotionally stable persons with
similar antisocial tendencies. In Eysenck's early work, E and N were therefore
linked to increased likelihood of antisocial and criminal behavior. As early as
1952, Eysenck had envisaged a personality trait P, which he later demonstrated
empirically to be an independent dimension from E and N. It was incorporated
into the EPQ in 1975, which measured Eysenck's three main dimensions of
160 Personality

personality, E, N, and P. Over the years Eysenck's personality questionnaire

has been extensively used to test his theory. His theory and the questionnaire
have been closely associated. Both have been refined and expanded in order to
incorporate new empirical findings from research conducted internationally.
Unlike E and N, P was not derived from theory. It was originally associated
more with schizophrenia and manic-depressive illness than with criminality,
but empirical evidence over the years has demonstrated its link with antisocial
and criminal behavior. There is a lack of theory linking P with socialization, but
in Eysenck's recent work he emphasizes the importance of conditioning in a
similar fashion to that which he originally did in the case of extraversion.
Whereas in Eysenck's early work E played a crucial role hi his theory of
criminality, P has now taken the more prominent role. A high P score is
associated with the severity and persistence of criminality, external attribution
of blame for their criminal offense, and it acts as a hindrance to psychological
therapies. In the future, attempts should be made to develop treatment
techniques that specifically focus on overcoming the attitudinal problems of
high P scorers. Cognitive therapy techniques may prove the most effective
treatment with these difficult to manage patients (Davidson & Tyrer, 1996).
The main problem with Eysenck's P dimension is that it is prone to false
negative error. That is, it sometimes fails to identify persistent and serious
offenders and the correlation with Hare's psychopathic checklist is very modest
(Hare, 1982). In addition, high P scorers only form a small part of the criminal
population (Blackburn, 1993). Therefore, a low P score may not be particularly
informative, whereas a high score is likely to indicate problems with
socialization. The problem may lie in the fact that the EPQ-R relies on self-
report. In the construction of the P dimension H. J. Eysenck and S. B. G.
Eysenck (1976) recognized the inherent difficulties in measuring this
dimension among insightless, hostile, and paranoid characters (see also
chapter 6 in this volume).
When conditions favorable to offending exist, such as being brought up in a
delinquent family and neighborhood, then a host of "protective factors" may
operate to counteract the risk of offending. These may involve high autonomic
arousal and conditioning and intellectual skills. The precise protective
mechanism is not fully understood, but it may involve, as Eysenck's theory
suggests, good conditionability facilitating the development of a conscience. As
far as prevention is concerned, the focus should be on identifying those
youngsters who are most at risk of becoming repeated offenders and providing
them with the necessary treatment and support (Farrington, 1991; Lynam,
1996). A small minority (about 6%) of offenders are responsible for about 50%
of all known crimes (Farrington, Ohlin, & Wilson, 1986; Wolfang, Figlio, &
Sellin, 1972). It therefore makes sense to understand exactly what predisposes
these individuals towards criminal activity and to find ways of effectively
dealing with them.
 Crime and personality 161

Eysenck's theory has been more influential in stimulating research into the
causes of crime than its prevention and control. Although his theory has some
implications for the prevention and control of criminality, it has not stimulated
much empirical research in this area. The major problem with much of the
research into the psychological treatment of offenders is the lack of theoretical
focus. Cognitive behavioral approaches appear to be most sound and effective
with some types of offenders (Eysenck & Gudjonsson, 1989). Applying Gray's
(1981) model of individual differences in sensitivity to rewards and punishment
may prove useful in determining the types of psychological approach that
should be provided for individual offenders. One of Eysenck's earliest
arguments was that matching treatment with the personality of the offender
will produce the best outcome results in terms of crime prevention. Individual
differences, both in terms of personality and the offenders' particular needs,
are undoubtedly of great importance when applying treatment programs with
certain types of offender.

REFERENCES

Berman, T., & Paisey, T. (1984). Personality in assaultive and non-assaultive juvenile
male offenders. Psychological Reports, 54, 527-530.
Blackburn, R. (1993). The psychology of criminal conduct. Theory, research and practice.
Chichester, U.K.: John Wiley & Sons.
Bradley, C., & Bowen, M. (1941). Amphetamine therapy of children's behavior
disorder. American Journal of Orthopsychiatry, 11, 92-103.
Cochrane, R. (1974). Crime and personality: Theory and evidence. Bulletin of the British
Psychological Society, 27, 19-22.
Cutts, K. K., & Jasper, H. H. (1939). Effect of benzedrine sulfate and phenobarbital on
behavior problem children with abnormal electroencephalograms. Archives of
Neurology and Psychiatry, 41, 1138.
Davidson, K. M., & Tyrer, P. (1996). Cognitive therapy for antisocial and borderline
personality disorders: Single case study series. British Journal of Clinical Psychology,
35, 413-429.
Douglas, J., & Olshaker, M. (1996) Mindhunter: Inside the FBI elite serial crime unit.
London: Heineman.
Enstrom, G., Alsen, M., Gustavsson, P., Schalling, D., & Traskman-Bendz, L. (1996).
Classification of suicide attempters by cluster analysis: A study of the temperamental
heterogeneity in suicidal patients. Personality and Individual Differences, 21, 687- 695,
Eysenck, H. J. (1952). The scientific basis of personality. London: Routledge & Kegan
Paul.
Eysenck, H. J. (1954). The psychology of politics. London: Routledge & Kegan Paul.
Eysenck, H. J. (1960). Symposium: The development of moral values in children. The
contribution of learning theory. British Journal of Educational Psychology, 30, 11-21.
Eysenck, (1964a). Crime and personality. London: Routledge & Kegal Paul.
162 Personality

Eysenck, H. J. (1964b). The biological basis of criminal behaviour. Advancement of

Science, September, 1-11.
Eysenck, H. J. (1964c). Biology and science. New Society, August.
Eysenck, H. J. (1964d). Biological factors in neurosis and crime. Kortiko-viszerale
physiologic, pathologic und therapie (pp. 169-175). Berlin: Akademie-Verlag.
Eysenck, (1970). Crime and personality. (2nd. ed.). London: Paladin Press.
Eysenck, (1977). Crime and personality. (Revised ed.). London: Routledge & Kegan
Paul Ltd.
Eysenck, H. J. (1995). Psychopathology: Type or trait? Behavioral and Brain Sciences,
18, 555-556.
Eysenck, H. J. (1996a). Personality and crime: Where do we stand? Psychology, Crime &
Law, 2, 143-152.
Eysenck, H. J. (1996b). Personality and the biosocial model of anti-social and criminal
behaviour. Paper presented at A Nato Advanced Study Institute on biosocial bases of
violence: Theory and research. Rhodes, Greece, 12-21 May 1996.
Eysenck, H. J., & Eysenck, S. B. G. (1970). Crime and personality: an empirical study of
the three factor theory. British Journal of Criminology, 10, 225-239.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality
Questionnaire (Junior and Adult). London: Hodder & Stoughton.
Eysenck, H. J., & Eysenck, S. B. G. (1976). Psychoticism as a dimension of personality.
London: Hodder and Stoughton.
Eysenck, H. J., & Gudjonsson, G. H. (1989). The causes and cures of criminality. New
York and London: Plenum Press.
Eysenck, S. B. G., Rust, J., & Eysenck, H. J. (1977). Personality and the classification of
adult offenders. British Journal of Criminology, 17, 169-179.
Farrington, D. P. (1989). Early predictors of adolescent aggression and adult violence.
Violence and Victims, 4, 79-100.
Farrington, D. P. (1991). Antisocial personality from childhood to adulthood. The
Psychologist, 4, 389-394.
Farrington, D. P., Biron, L., & LeBlanc, M. (1982). Personality and delinquency in
London and Montreal. In: J. Gunn & D. P. Farrington (Eds.), Abnormal offenders
and the criminal justice system (pp. 121-148). New York: John Wiley.
Farrington, D. P., Ohlin, L. E., & Wilson, J. Q. (1986). Understanding and controlling
crane: Towards a new research strategy. New York: Springer-Verlag.
Fonseca, A. C, & Yule, W. (1995). Personality and antisocial behavior in children and
adolescents: An inquiry into Eysenck's and Gray's theories. Journal of Abnormal
Child Psychology, 23, 767-781.
Franks, C. M., & Trouton, S. S. (1958). Journal of Comparative and Physiological
Psychology, 51, 221.
Gabrys, J. B., Peters, K., Robertson, G., Utendale, K. A., Schumph, D., Laye, R. C.,
O'Haire, T., Allard, I., & Phillips, N. (1988). Personality attributes of children with
conduct disorders: The discriminative power of the Junior EPQ. Psychological
Reports, 62, 63-69.
Goma-i-Freixanet, M. (1995). Prosocial and antisocial aspects of personality. Personality
and Individual Differences, 19, 125-134.
Gough, H. G. (1960). Theory and measurement of socialization. Journal of Consulting
Psychology, 24, 23-30.
 Crime and personality 163

Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.),

A model of personality (pp. 246—276). New York: Springer.
Gudjonsson, G. H. (1984). Attribution of blame for criminal acts and its relationship
with personality. Personality and Individual Differences, 5, 53-58.
Gudjonsson, G. H. (1987). The significance of depression in the mechanism of
"compulsive" shoplifting. Medicine, Science and the Law, 27, 171-176.
Gudjonsson, G. H., Petursson, H., Sigurdardottir, H., & Skulason, S. (1991). The
personality of Icelandic prisoners: some normative data. Nordic PsyJdatr. Tidsskr., 45,
151-157.
Gudjonsson, G. H., & Roberts, J. C. (1983). Guilt and self-concept in "secondary
psychopaths." Personality and Individual Differences, 4, 65-70.
Gudjonsson, G. H., & Roberts, J. C. (1985). Psychological and physiological
characteristics of personality-disordered patients. In D. P. Farrington, & J. Gunn
(Eds.), Aggression and dangerousness (pp. 81-101). Chichester, U.K.: John Wiley &
Sons.
Gudjonsson, G. H., & Singh, K. K. (1989). The revised Gudjonsson Blame Attribution
Inventory. Personality and Individual Differences, 10, 67-70.
Hare, R. D. (1982). Psychopathy and the personality dimensions of psychoticism,
extraversion and neuroticism. Personality and Individual Differences, 3, 35-42.
Holmes, R. M., & De Burger, J. (1988). Serial murder. London: Sage Publications.
Howarth, E. (1986). What does Eysenck's Psychoticism scale really measure? British
Journal of Psychology, 77,223-227.
Kendel, E., Mednick, S., Kirkegaard-Sorensen., L., Hutchings, B., Rosenberg, R., &
Schulsinger, F. (1988). I.Q. as potential factor for subjects at high risk for antisocial
behavior. Journal of Consulting and Clinical Psychology, 56, 224-226.
Lane, D. A. (1987). Personality and antisocial behaviour: A long-term study. Personality
and Individual Differences, 8, 799-806.
Lindsley, D. D., & Henry, C. E. (1942). Psychosomatic Medicine, 4, 140.
Lykken, D. (1957). A study of anxiety in the sociopathic personality. Journal of
Abnormal and Social Psychology, 55, 6-10.
Lynam, D. R. (1996). Early identification of chronic offenders: Who is the fledgling
psychopath? Psychological Bulletin, 120, 209-234.
McEwan, A. W. (1983). Eysenck's theory of criminality and the personality types and
offences of young delinquents. Personality and Individual Differences. 4, 201-204.
McGurk, B. J., & McDougall, C. (1981). A new approach to Eysenck's theory of
criminality. Personality and Individual Differences, 2, 338-340.
Mealey, L. (1995). The sociobiology of sociopathy: An integrated evolutionary model.
Behavioral and Brain Sciences, 18, 523-599.
Moir, A., & Jessel, D. (1995). A mind to crime. The controversial link between the mind
and criminal behaviour. London: Michael Joseph.
Mosher, D. L. (1966). The development of multitrait-multimethod matrix analysis of
three measures of three aspects of guilt. Journal of Consulting Psychology, 30, 25-29.
Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning.
Urbana, IL: University of Illinois Press.
Passingham, R. E. (1967). Crime and personality: A review of Eysenck's theory.
Unpublished master's thesis, University of London.
164 Personality

Passingham, R. E. (1972). Crime and personality: A review of Eysenck's theory. In V.

D. Nebylitsin, & 3. A. Gray (Eds.), Biological bases of individual behaviour. New
York: Academic Press.
Patterson, G. R. (1986). Performance models for antisocial boys. American Psychologist,
41, 432-444.
Powell, G. E. (1977). Psychoticism and social deviancy in children. Advanced Behaviour
Research & Therapy, 1, 27-56.
Rachman, M. A., & Eysenck, S. B. G. (1978). Psychoticism and response to treatment in
neurotic patients. Behaviour Research and Therapy, 16, 183-189.
Raine, A. (1993). The psychopathology of crime. New York: Academic Press.
Raine, A., & Venables, P. (1981). Classical conditioning and socialization: A bisocial
interaction. Personality and Individual Differences, 2, 273-283.
Raine, A., Venables, P., & Williams (1995). High autonomic arousal and electrodermal
orienting at age 15 years as protective factors against criminal behavior at age 29
years. American Journal of Psychiatry, 152, 1595-1600.
Raine, A., Venables, P., & Williams (1996). Better autonomic conditioning and faster
electrodermal half-recovery time at age 15 years as possible protective factors against
crime at the age of 29 years. Developmental Psychology, 32, 624-630.
Rushton, J. P., & Christjohn, R. D. (1981). Extraversion, neuroticism, psychoticism and
self-reported delinquency: Evidence from eight separate samples. Personality and
Individual Differences, 2, 11-20.
Salter, A. (1988). Treating child sex offenders and victims. London: Sage Publications.
Schalling, D. (1978). Psychopathy-related personality variables and the psychophysiol-
ogy of socialization. In R. D. Hare, & D. Schalling (Eds.), Psychopathic behaviour
Approaches to research (pp. 85-106). Chichester, U.K.: John Wiley & Sons.
Schuessler, K. F., & Cressey, D. R. (1950). Personality characteristics of criminals.
American Journal of Sociology, 55, 476-484.
Sears, D. J. (1991). To kill again. The motivation and development of serial murder.
Wilmington, DE: Scholarly Resources Inc.
Solomon, R. L., Turner, L. H., & Lessac, M. S. (1968). Some effects of delay of
punishment on resistance to temptation in dogs. Journal of Personality and Social
Psychology, 8, 233-236.
Trasler, G. B. (1978). Relations between psychopathy and persistent criminality:
Methodological and theoretical issues. In R. D. Hare, & D. Schalling (Eds.),
Psychopathic behavior: Approaches to research (pp. 273-298). New York: John Wiley
& Sons.
Trasler, G. B. (1979). Delinquency, recidivism and desistance. British Journal of
Criminology, 19, 314-322.
Wilson, G. D., & Cox, D. N. (1983). Personality of paedophile club members.
Personality and Individual Differences, 3, 323-329.
Wolfang, M. E., Figlio, R. M., & Sellin, T. (1972). Delinquency in a birth cohort.
Chicago, IL: University of Chicago Press.
Zuckerman, M. (1989). Personality in the third dimension: A psychobiological
approach. Personality and Individual Differences, 10, 391-418.
Chapter 9
Sex and personality

G. D. Wilson

1. INTRODUCTION

Hans Eysenck's main contribution to sex research was to note and fill a gap in
the celebrated work of Kinsey and Masters and Johnson. These researchers
had concentrated on identifying typical patterns of sexual behavior, and even
though they had observed enormous individual differences in passing, they
confined themselves to comparisons only involving major demographic varia-
bles such as age, gender, and social class. Eysenck therefore determined to
extend sex research by investigating the links between constitutional
personality factors and sexual attitudes and behavior. Most of this research
is summarized in his 1976 book Sex and Personality.
Some literature on this question was already available but it was scant.
Schofield (1968) had described the characteristics of the most sexually active
young people in Britain, and although no questionnaire measure was used they
fitted the description of extraverts (outgoing, sociable, and generally active).
One of the best single indicators of sexual experience, especially for girls, was
cigarette smoking (also known to go with E). There was also the work of Giese
and Schmidt (1968) in Germany, indicating that extraverts engaged more than
introverts in virtually every form of sexual activity except masturbation (which
of course is a solitary activity), a finding confirmed by Husted and Edwards
(1976). Giese and Schmidt also found that high N went with higher rates of
masturbation, greater desire for intercourse, and more spontaneous erections
in men, and with less frequent orgasms and greater menstrual discomfort in
women. P was not well recognized as a major dimension at this time, although
Zuckerman and colleagues (1974, 1976) had found that all kinds of sexual
experience were more common in sensation seekers.

2. EPQ CORRELATES OF SEXUAL BEHAVIOR

For assessing sexual behavior and attitudes, Eysenck assembled a question-

naire containing over 100 items. This was derived partly from existing
166 Personality

questionnaires, such as that of Thome, Haupt, and Allen (1966), but was
supplemented by contributions from members of Hans's Department such as
Maurice Yaffe and myself, whom he considered streetwise. This covered a
wide range of sexual preferences and difficulties; examples of the items are
given in Table 9.1.
The questionnaire was initially completed by 423 male and 379 female
unmarried students, aged 18-22, from various colleges around Britain, who
also completed the EPQ. Answers to the sex questionnaire were analyzed by
principal components with promax rotation to yield 14 primary factors that
were readily interpretable and similar for men and women (Table 9.1). Since
these were oblique they were refactored to reveal two orthogonal higher order
factors called sexual pathology and libido. The former included all kinds of

Table 9.1. Sexual attitude areas identified by factor analysis (from Eysenck, 1976)

Factors Examples of items

1. Satisfaction I have not been deprived sexually
My love life has not been disappointing
2. Excitement It doesn't take much to get me excited sexually
I get very excited when touching a woman's breasts
3. Nervousness I don't have many friends of the opposite sex
I feel nervous with the opposite sex
4. Curiosity Sex jokes don't disgust me
I would agree to see a "blue" film
5. Premarital sex Virginity is a girl's most valuable possession
One should not experiment with sex before marriage
6. Repression Children should not be taught about sex
I think only rarely about sex
7. Prudishness I don't enjoy petting
The thought of a sex orgy is disgusting to me
8. Experimentation A person should learn about sex gradually by experimenting with it
Young people should be allowed out at night without being too closely
checked
9. Homosexuality I understand homosexuals
People of my own sex frequently attract me
10. Censorship There are too many immoral plays on T.V.
Prostitution should not be legally permitted
11. Promiscuity Sex without love ("impersonal sex") is not highly unsatisfactory
I have been involved in more than one sex affair at the same time
12. Hostility I have felt like humiliating my sex partner
I have felt hostile to my sex partner sexually
13. Guilt At times I have been afraid of myself for what I might do sexually
My conscience bothers me too much
14. Inhibition My parents' influence has inhibited me sexually
Conditions have to be just right to get me excited sexually
 Sex and personality 167

difficulty, complaint, and peculiarity, while the latter combined active, intense
sexuality with permissive attitudes. These superfactors were also fairly similar
for men and women and were virtually uncorrelated within each gender.
When EPQ dimensions were projected onto these sex factors the results
shown in Table 9.2 were obtained. Extraverts were high on Promiscuity and
low on Nervousness and Prudishness ("happy philanderers"); introverts
tended to be more puritanical, valuing virginity and fidelity and playing down
the importance of physical sex. High N scorers were high on Excitement,
Nervousness, Guilt, and Inhibition and were low on Satisfaction. High P
scorers were high on Curiosity, Premarital sex, Promiscuity, and Hostility.
These patterns were seen as consistent with predictions based on the general
nature of the these personality types, with neurotics being highly driven but
inhibited by conscience (anxiety, guilt, etc.) and psychotics acting out libidi-
nous, promiscuous, and sometimes antisocial, desires with little regard for the
feelings of other people. Questions concerning sexual problems revealed that
N was connected with orgasm and erectile difficulties while neither E nor P
were much implicated in dysfunction.
The Lie scale was also correlated with sexual behavior, high L scorers
appearing as somewhat unadventurous and conventionally well-behaved (like
low P scorers in many respects). This could mean either that they were "faking
good" on the sex questionnaire, or that they actually are conforming, "resp-
ectable" people. The latter interpretation is supported by research showing
that conservatism (which correlates with L) is associated with restricted sexual

Table 9.2. Sexual attitudes related to personality (from Eysenck, 1976)

Factors E N
1. Satisfaction +
2. Excitement + ++
3. Nervousness ++
4. Curiosity 0 +
5. Premarital sex + 0
6. Repression 0 0
7. Prudishness — +
8. Experimentation + 0
9. Promosexuality 0 +
10. Censorship - 0
11. Promiscuity ++ 0
12. Hostility 0
13. Guilt 0
14. Inhibition 0
Summary factors
A. Sexual pathology —
B. Libido +
Note. +, 0, — signs indicate positive, zero, and negative relationships, respectively.
168 Personality

experience and a general dislike of sexual stimuli (Joe, Brown, & Jones, 1976;
Schmidt, Sigusch, & Meyberg, 1969; Thomas, 1975; Thomas, Shea, & Rigby,
1971). Farkas, Sine, and Evans (1979) found penile tumescence responses to
erotic stimuli to relate inversely to L scores, which they also interpreted as
being mediated by conservative attitudes. (Other EPQ variables showed no
relationship with physiological arousal except for a weak relationship between
N and speed of attaining maximum tumescence. Griffith and Walker (1975)
had previously failed to find any relationship between E and N and arousal
ratings of erotic slides.)
The results so far described were obtained with student samples. Eysenck
later went on to study a more general adult sample of 427 males and 436
females. Results were fairly similar. Factor analysis produced a similar set of
primary factors arranged in relation to two orthogonal higher order factors,
this time labeled Libido and Satisfaction (Figure 9.1). Extraversion was again
associated with permissiveness, strong libido and desire for variety in partners
and activities, N went with excitability and a wide range of difficulties and
conflicts, and P went with a tough, adventurous and impersonal approach to
matters of love and sex. Again, L scores were related to conservative attitudes
and behavior.
Since Eysenck's studies there has been little work on the relationship
between personality traits and sexual behavior. The nearest to replication were
the German studies of Schenk and Pfrang (1986) who found that three
markers of active, variform sexuality (age of first intercourse, number of
partners, and frequency of intercourse) went with extraversion only in young,
unmarried men. N was not related and P was not measured (although a
measure of "acting out" aggression did go with all three indicators of intense
sexuality). Schenk and Pfrang suggested that personality correlates are
obscured in married people by the quality of the marital relationship.
However, their studies were rather limited with respect to the sexual variables
measured.

3. MALE-FEMALE DIFFERENCES

Eysenck also analyzed his sex questionnaire to identify items that discriminated
men and women, producing an empirically derived "masculinity-femininity
scale" (Table 9.3). Males were revealed as more favorable to pornography,
promiscuity, voyeurism, prostitution, premarital, and impersonal sex than
females. They were also more easily aroused sexually and masturbated more
often. But although men were more permissive and felt less guilt, women
expressed greater contentment with their sex lives. This pattern of findings
could be explained by supposing that men have a stronger libido than women
(especially as regards desire for variety in partners) with the result that women
 Sex and personality 169

-l.U
2
* 0.
(2) Satisfaction
• Males
O Females

•0.6

-0.4

- (11) Physical sex W Sexual excitement

-0.2
11 (1 ) Permissiveness
• O (5) Pornography
-0.6 -0.4 -0.2 0.2 0.4 0.6 5 o 1.0
i i i i i i i i i i i i i* i i i
9 (12) Aggressive sex «1 LIBIDO
12 O (4) Impersonal sex
(7) Prudishness » 9 - -0.2 •4
o
•7
O
(9) Sexual disgust o --0.4
(6) Sexual shyness
AR
o (3) Neurotic sex
L-0.6 «3
SATISFACTION

Figure 9.1. Two major dimensions of sexual attitudes and behaviour (from Eysenck, 1976).

command a "sellers' market" and men are thus more often frustrated in their
desires. Such an interpretation is supported by Wilson's (1978) finding of a
greater discrepancy between sexual fantasy and reality in men than in women,
and by Zuckerman et al. (1976) who found a higher correlation between sexual
attitudes and activities in women than in men. More recently it has become
clear that the "higher libido" of males refers to a desire for variety in partners
and activities rather than preferred frequency of intercourse within a loving
partnership (Wilson, 1989).
The male-female (M-F) differences in Table 9.3 are similar to within-sex
differences between high and low P individuals. In other words, P seems to
parallel masculinity as regards its effects on sexual inclinations. Correlations
between P and M-F scores were .54 and .74 for men and women, respectively.
(There was also a significant, albeit much lower association between E and
170 Personality

Table 93. Sexual attitudes that differentiate men and women (from Eysenck & Wilson, 1979)

M F Difference
1. Sex without love ("impersonal sex") is highly unsatisfactory 43 60 -17
2. Conditions have to be just right to get me excited sexually 15 42 -27
3. Sometimes it has been a problem to control my sex feelings 50 38 + 12
4. 1 get pleasant feelings from touching my sexual parts 81 66 + 15
5. 1 do not need to respect a sex partner or love him/her, in 43 26 + 17
order to enjoy petting and/or intercourse with him/her
6. Sexual feelings are sometimes unpleasant to me 6 11 -5
7. It doesn't take much to get me excited sexually 75 44 +31
8. 1 think about sex almost every day 87 61 +26
9. 1 get excited sexually very easily 68 40 +28
10. The thought of a sex orgy is disgusting to me 15 40 -25
11. 1 find the thought of a colored sex partner particularly 32 11 +21
exciting
12. 1 like to look at sexy pictures 80 45 +35
13. My conscience bothers me too much 13 20 _7
14. 1 enjoy petting 95 88 +7
15. Seeing a person nude doesn't interest me 6 28 -22
16. Sometimes the woman should be sexually aggressive 95 88 +7
17. 1 believe in taking my pleasures where I find them 34 19 + 15
18. Young people should be allowed out at night without being 69 54 + 15
too closely checked
19. 1 would particularly protect my children from contact with sex 6 12 -6
20. 1 like to look to pictures of nudes 84 44 +40
21. If I had the chance to see people making love, without being 67 37 +30
seen, I would take it
22. Pornographic writings should be freely allowed to be 74 55 + 19
published
23. Prostitution should be legally permitted 82 63 + 19
24. 1 had some bad sex experiences when I was young 13 20 -7
25. There should be no censorship, on sexual grounds, of plays 73 53 +20
and films
26. Sex is far and away my greatest pleasure 35 26 +9
27. Absolute faithfulness to one partner throughout life is nearly 41 28 + 13
as silly as celibacy
28. The present preoccupation with sex in our society has been 45 54 -9
largely created by films, newspapers, television and
advertising
29. 1 would enjoy watching my usual partner having intercourse 18 6 + 12
with someone else
30. 1 would vote for a law that permitted polygamy 31 11 +20
31. Even though one is having regular intercourse, masturbation 55 39 + 16
is good for a change
32. 1 would prefer to have a new sex partner every night 7 2 +5
33. Sex is more exciting with a stranger 21 7 + 14
34. To me few things are more important than sex 44 26 + 18
35. Sex is not all that important to me 11 19 -8
36. Group sex appeals to me 33 10 +23
37. The thought of an illicit relationship excites me 52 32 +20
38. 1 prefer my partner to dictate the rules of the sexual game 9 37 -28
39. The idea of "wife swapping" is extremely distasteful to me 37 63 -26
40. Some forms of love-making are disgusting to me 15 30 -15

Note. Percentage endorsements of men and women and the difference between them.
 Sex and personality 111

masculinity.) Sex offenders are also known to be high on P; the mean P for sex
offenders (including those convicted of rape, indecent assault, and buggery)
was 11.07 compared with a prisoner mean of 6.25. Thus active and antisocial
sexuality seems to correlate with both P and masculinity, male hormones
presumably being the mediating factor.
Within Eysenck's adult sample, an interesting gender difference appeared:
participation in group sex went with high E in men, but high P in women. This
could mean that a liking for group sex is so "normal" for men that it simply
reflects their sociability, whereas women who engage in such behavior are
"unfeminine," hence unusual, and more likely to be eccentric and non-
conforming (high on P). Perhaps for similar reasons, high levels of P-like traits
have been reported for prostitutes (O'Sullivan, Zuckerman, & Kraft, 1996) and
sadomasochistic women (Gosselin, Wilson, & Barrett, 1991).
In passing we might note that gender differences in personality itself have
been firmly established and shown to have considerable cross-cultural
consistency (Lynn & Martin, 1997). Females tend to score higher on N,
males on P, and there is no clear difference with respect to E. The differences
in N and P are presumed to have a brain basis mediated by hormones (Ellis,
1986).

4. CRIMINALITY AND DEVIANCE

Eysenck (1976) gave his sex questionnaire to a sample of 186 patients in

Broadmoor Hospital for the criminally insane. Most of these patients had
committed crimes of violence such as murder and assault but there were also
some sex offenders and arsonists. The outstanding feature of the results was
that the patients appeared sexually inhibited compared with normal controls.
They claimed not to be readily aroused, to think rarely about sex, and to look
on it as being for procreation rather than pleasure. This was consistent with a
finding of Wilson and Maclean (1974) that, compared with bus drivers, male
prisoners were less favorable towards sexual freedom and less inclined to laugh
at sexual jokes. Similarly, Thorne et al. (1966) reported felons and sex
offenders in the U.S. to be repressed and conservative in sexual attitudes.
Of course prisoners may be faking good on the questionnaires in the hope of
earlier release, and indeed their EPQ Lie scores are elevated (8.03 for
Broadmoor patients vs. 3.64 for normals). However, accounts of their actual
sexual experience also suggest a restrictive background; there were more things
they would like to have done and more things they had done that they did not
enjoy. On their own report they seem more inhibited and less satisfied. Again,
the fact of their incarceration may be responsible for some lack of experience
and dissatisfaction. The above studies do not permit the untangling of these
factors.
172 Personality

Wilson and Gosselin (1980) studied the personality of sexually variant men
in the community by obtaining samples from the membership of clubs for
fetishists, transvestites, transsexuals, and sadomasochists and from the mailing
lists of suppliers of special garments and equipment for people with such
predilections. Results (Table 9.4) indicated that variant men were generally
rather introverted and high on N compared with age-matched male controls;
their N scores were, however, no higher than normal women. P scores of
variant men were not significantly elevated, but a group of "dominant" women
who specialized in catering for their preferences were significantly elevated on
P, as well as being rather extraverted. The tendency for variant men to be shy
and introverted was confirmed in a study of paedophiles by Wilson and Cox
(1983) while Gosselin et al. (1991) found sadomasochistic women to be high on
P and E and low on N (as well as being very active sexually). Thus it appears
that variant men and women are virtually opposite in personality and hence the
perfect foil for each other.
In seeking to explain the personality characteristics of variant men, Wilson
and Gosselin (1980) supposed that a lack of social skill and confidence
(introversion) or a tendency to be made anxious by women (high N) might
have caused them to gravitate towards impersonal sex targets such as rubber
and women's clothing. However, it is also possible that self-consciousness
concerning their unusual sexual predilection leads to their social isolation—the
belief that partners would react badly to the discovery of the variant
preference. Another possible factor to be considered is the greater
conditionability of introverts, which might increase the likelihood of the
accidental association of sexual arousal with inanimate objects; this would help
to explain the positive enjoyment that many variant men obtain from their

Table 9.4. Mean scores (and standard deviations) of variant and normal groups on personality
factors measured by the Eysenck Personality Questionnaire (from Wilson & Gosselin, 1980)

No. in Lie
group Extraversion Neuroticism Psychoticism scale
Sadomasochists 133 10.68 (5.3) 11.03 (5.9) 3.50 (3.0) 7.92 (4.2)
Rubberites 87 10.87 (-) 10.49 (-) -(-) 8.91 (-)
Leatherites 38 12.55 (4.6) 9.94 (5.9) 3.29 (1.9) 8.97 (4.6)
Transvestites 269 9.96 (5.4) 13.27 (5.8) 3.27 (2.3) 8.32 (4.2)
Transsexuals 16 9.50 (5.2) 13.06 (5.5) 3.19 (2.3) 11.12 (4.2)
Dominant women 25 14.04 (5.5) 12.68 (4.5) 5.92 (3.8) 7.00 (4.4)
Control males3 50 12.38 (5.1) 9.17 (5.1) 3.09 (2.6) 8.07 (4.1)
Control femalesb 416 12.24 (4.9) 12.63 (5.4) 2.35 (2.1) 8.86 (3.9)
Control females0 27 11.97 (5.0) 12.57 (5.3) 2.28 (2.2) 8.84 (4.1)
"Control males are matched in age and social class with male variant groups.
b
This group of female was age-matched with the various male groups from EPiQ standardization
data.
°These control females were collected specifically for this study and were age-matched for
comparison with the dominant women.
 Sex and personality 173

fixations. Once acquired, the variant interests might incubate because they
isolate the individual further from social contacts. Transvestites might be an
exception to this rule, however: Gosselin and Eysenck (1980) found that when
cross-dressed and playing the female role they were more relaxed and sociable
than when in the male mode (this observation being supported by a shift in
EPQ scores in the direction of lowered N and increased E).
Gosselin and Wilson (1980) also found that personality scores could be used
to distinguish one sexual variation from another. When sadomasochists were
divided into those who were primarily sadistic and those whose main interest
was in masochism, the masochists appeared as more introverted than the
sadists, and slightly lower on P. In fact, neither the sadists nor the leatherites
were significantly different from control males on E, while all the other variant
male groups were significantly introverted.
Homosexual and bisexual men also tend towards introversion and high N
(Evans, 1970; Wilson & Fulford, 1979), as do lesbians (Eisinger et al., 1972).
However, the personality differences are not very striking, and the high N
could be due to the fact that Ss were often obtained from clinical sources. In
today's more permissive social climate there might be less anxiety arising from
the marginal status of homosexuality, so the differences may have disappeared.
Wilson and Fulford found that bisexual men were higher in P than either
exclusive homosexuals or heterosexual men, one of several observations that
led them to conclude that bisexuals were characterized by high levels of libido
and masculinity, leading to greater generalization of sex targets.

5. SEXUAL DYSFUNCTION

The personality correlates of female orgasm difficulty are unclear. Eysenck

(1976) found orgasm problems to be associated with high N, but Bentler and
Peeler (1979) found virtually no relationship and Shope (1968) found the
opposite—orgasmic women were less stable than nonorgasmic women.
Unfortunately, Shope gave few details of his measure of "emotional stability"
so it may have been nearer to "expressiveness" than N (i.e., an extravert
characteristic). Fisher (1973) reported that orgasmic women were distin-
guished in voice quality, being judged as more natural and variable in tonal
range, with more use of emphatic and dramatic sounds like sighs and deep
breaths. These attributes also suggest expressiveness and approximate to the
vocal sounds that often accompany women's orgasm. While it might seem
logical that high N would be associated with orgasm difficulty, state anxiety
does not seem counter to orgasmic capacity since anxiolytic drugs are not an
effective treatment for orgasmic dysfunction (Wilson, 1987) and female sexual
arousal may be facilitated by fear (Medicus & Hopf, 1990). Bentler and Peeler
(1979) found a tendency for extraverted women to be more orgasmic but felt
174 Personality

that this could be accounted for by their higher levels of sexual experience
rather than any more fundamental neurological factors. Vaginismus does seem
to be associated with anxiety and high N, as well as low self-esteem, though the
direction of cause and effect is arguable (Kennedy, Doherty, & Barnes, 1995).
Finally, a study of women with histrionic personality disorder (high E and N?)
found them to be relatively "erotophobic," anorgasmic, and dissatified with
their marriage (Apt & Hurlbert, 1994).
High P might be expected to be associated with orgasm, since high P women
are male-like and men seldom have difficulty in this regard. Testosterone
injections make women more sexually responsive (Kane, Lipton, & Ewing,
1969) and persistence (another typically male attribute) is also correlated with
orgasmic ability (Fisher, 1973). It is thus surprising that Eysenck's research
failed to show any significant association between P and orgasmic functioning
(although the trend was in the right direction).
Eysenck's finding of high N in men with potency disorders (mentioned
above) was complemented by Munjack, Kanno, and Oziel (1978) who found
high levels of anxiety, depression, and general psychopathology in men with
ejaculatory problems (both premature and retarded forms). These patients
were also higher than controls on the Schizophrenia and Social Introversion
scales of the MMPI, suggestive of high P and low E. The only variable to
distinguish premature from retarded ejaculation was the Masculinity-
Femininity scale of the MMPI, retarded ejaculation being more "masculine."
The interpretation of this finding is unclear.

6. MARITAL CHOICE AND SATISFACTION

Many of Eysenck's subjects were married couples, so it was possible to examine

their similarity with respect to personality. A slight similarity effect emerged
(correlations being P .14, E .06, N .22, L .17). Given the large sample, all these
are significant except for E. Taking these together with more recent results
(Eysenck & Wakefield, 1981; Insel, 1971; Nias, 1977), low positive correlations
on EPQ variables in married couples are indicated, those for P and N being
higher than that for E (which approximates to zero). Such similarity that does
appear seems to result from initial partner choice rather than progressive
merging of personalities through the course of the marriage. This is surprising
considering that Bentler and Newcomb (1979) found a higher divorce rate
among couples who were initially dissimilar in personality, which should result
in a tendency for couples in long-standing marriages to appear as more similar.
Eysenck and Wakefield actually found a slight tendency for the libidos of
husband and wife to diverge with time into the marriage. Wives of high libido
men showed a greater decrement in libido with age than wives of low libido
 Sex and personality 175

men. However, the data were cross-sectional not longitudinal, so caution

should be exercised in interpretation.
Other attributes show much higher correlations between husband and wife.
Eysenck and Wakefield (1981) found correlations of .73 for marital
satisfaction, .41 for sexual satisfaction, .43 for libido, .51 for radicalism, and
.56 for tendermindedness. Similar correlations in married couples have been
reported for conservatism (.53), dogmatism (.51), and inflexibility (.41) by
Kirton (1977). These levels are similar to the degree of assortative mating
typically observed for IQ (Eysenck, 1979), although some of the apparent
homogamy in attitudes may be eliminated by partialing out age.
Catlin, Croake, and Keller (1976) found that cohabiting couples differed
from married couples by being slightly higher on the Psychopathic Deviate and
Hypomania scales of the MMPI for both sexes and the Masculinity and
Schizophrenia scales for men only. Such a pattern suggests high P in co-
habiters, but as cohabitation becomes increasingly common its nonconformity
aspect may diminish. In Sweden, at a time when half of all young couples were
cohabiting, little personality difference was found between cohabitors and
married couples.
Eysenck and Wakefield (1981) studied the relationships between
personality, attitudes, sexual behavior, and marital satisfaction in a sample of
566 couples recruited through newspaper and magazine advertisements. These
couples had been married for varying lengths of time, ranging from 0 to 40
years and their compatibility was assessed with an extended form of the Locke-
Wallace Marital Adjustment Test. First to be analyzed was the role of
personality in predicting marital satisfaction in themselves and their partner.
Consistent with earlier studies (Eysenck, 1980; Zaleski & Galkowska, 1978),
high P and N were associated with marital unhappiness in both self and spouse,
while E had little effect. The Lie score showed a small but significant
correlation with satisfaction for the testee only (not their spouse), which might
be seen as supporting its validity as an index of social desirability bias.
Combining the personality scores of husband and wife, a multiple correlation
of .43 was obtained with total marital satisfaction (the sum of MS for husband
and wife), with male and female scores contributing about equally to the
predictive power of personality.
Eysenck and Wakefield (1981) were also interested in the way in which
marital satisfaction could be predicted by the personality differential of the two
parties, bearing on the old dispute between similarity and complementation
theories of marital compatibility. Eysenck and Wilson (1979) had earlier
proposed a compromise position called gender-asymmetry theory, according to
which marital satisfaction is greatest when sex-typical differences (regardless of
whether they are due to biology or culture) are matched within the individual
couple—otherwise the similarity principle is presumed to prevail. A simple
example is height, where tall men marry women who are also tall but not as tall
176 Personality

as themselves, whereas short men many even shorter women. The result is a
high correlation (similarity, or assortative mating) but a predictable average
difference of 4-5 in. Gross deviations in either direction are presumed to be
unstable at all stages of a relationship from first meeting, through courtship to
the later years of marriage. Since men are typically higher on P than women,
and women typically higher on N, it would follow that higher levels of P in
husbands and higher levels of N in wives should be better tolerated than
differences which run counter to this expectation.
Evidence that generally supported this theory was found by Eysenck and
Wakefield (1981). Besides the generally detrimental effect of P as described
above, it was better for the husband if his wife had a similar P score and better
for the wife if the husband's P score was about 1.5 points higher than her own
(this being approximately equivalent to the average difference in P scores
between men and women). In the case of N, satisfaction was optimal for the
husband when his wife was 3.25 points higher than himself (again representing
the typical M-F difference), while for wives satisfaction was optimized when
the husband was about the same on N (see Figure 9.2).
When all the effects of personality were considered (both individual scores
and interactions between partners), 20% of the variance in marital satisfaction
was accounted for. The L score contributed about 1% of this, and N and P
nearly all the rest.
Eysenck and Wakefield (1981) also used a social attitudes inventory in their
study and found that radicalism was associated with dissatisfaction in both self
and spouse, whereas tendermindedness did not predict. However, when
partner differentials were examined it was tendermindedness rather than

Males MS Female MS

MS

points -3.37-3.25 -0.31 0

Female higher (No difference) Male higher

H

(actual mean difference) NEUROTICISM DIFFERENCE

Figure 92. Relationship between neuroticism differences and marital satisfaction (MS) of males
and females (from Eysenck & Wakefield, 1981).
 Sex and personality 177

radicalism that emerged as important. Husbands were happiest with a wife 3.5
points more tenderminded than themselves, while wives thrived on a husband 6
points tougher (again reflecting the average M-F difference on the scale).
These results for toughmindedness are similar to those for P, which is not
surprising considering the conceptual and empirical overlap of the two scales.
The Sexual Attitudes Inventory was also included in Eysenck and Wake-
field's study and not surprisingly sexual satisfaction correlated with overall
marital satisfaction. More interesting was the finding that high libido in men
was counter to marital satisfaction, whereas female libido was unrelated. In
libido also, sex-typical differences were associated with satisfaction, men being
happiest when 2.49 points higher on libido than their wives, and women being
most satisfied with a husband 4.94 points higher in libido than themselves
(Figure 9.3). Eysenck and Wakefield note that since the average M-F libido
difference was 12.19 points, most men would be too much higher than their
wives for optimum compatibility, which might explain why low libido men
appeared more content with their marriage.
Various other interesting sex differences appeared in the Eysenck and
Wakefield (1981) study. For example, sexual satisfaction was more integral to
the overall marital satisfaction of wives than husbands. Men readily attained
their optimal sexual satisfaction but women seldom did, a fact which might be
involved in many a marriage breakdown. A wife's report of impotence in her
husband was associated with unhappiness in both spouses but the husband's
awareness of nonorgasm in his wife did not diminish his own satisfaction to any
appreciable extent. Experience of premarital intercourse made no difference to
the marital satisfaction of the husband, but the same in the wife predicted
discontent.

Female MS

MS

I I \ I
0 2.49 4.49 12.94

Female higher (No difference) Male higher

H (+)

LIBIDO DIFFERENCE (actual mean difference)

Figure 9.3. Relationship between libido differences and marital satisfaction (MS) of males and
females (from Eysenck & Wakefield, 1981).
178 Personality

Many other facts could be gleaned from this excellent study but the main
finding was that personality, attitudes, and libido have effects upon marital
happiness both separately and in combination. Similarity is upheld, but must
be understood in the context of gender-asymmetry complementation, that is,
couples who replicate sex-typical differences in personality or other attributes
are likely to be happier than those running counter to stereotype. It also seems
to be the case that having both partners high on an attribute (such as P) which
is detrimental to marital happiness is usually worse than having only one
partner high on that attribute.
More recent studies of the personality predictors of marital happiness have
generally supported Eysenck's findings. Russell and Wells (1991) found low
levels of assortative mating for EPQ dimensions in 94 married couples (Lie .37,
P .28, N .18, E .05). Differences between partners in E, P, and L were
associated with worse marriages but differences in N did not matter.
Unfortunately, these authors did not consider the direction of the difference
in order to examine the gender-asymmetry hypothesis. They did, however,
report that the degree to which similarity on an item predicted marital
satisfaction was related to the heritability of that item, providing support for a
genetic similarity view of compatibility (Rushton, Russell, & Wells, 1984). In a
subsequent study, Russell and Wells (1994) examined the personality
predictors of marital happiness using a sample of 1200 couples and some
selected questions to tap extraversion and neuroticism. Husband's E was
associated with greater happiness in both partners, but wife's E was
immaterial. Neuroticism was detrimental to happiness in both husbands and
wives, that of the wife being much more significant. Unfortunately, P was not
included in this study, nor were the personality interactions between husband
and wife. Russell and Wells speculated that because personality similarity
generally makes for better marriages "it could be that a pair of neurotic
spouses are not as unhappy as an additive model would predict." The work of
Eysenck and Wakefield, however, suggests otherwise.
Other studies of personality in relation to marital happiness have used the
MMPI (Lucas & Peterson, 1991; Richard, Wakefield, & Lewak, 1990). These
confirm that personality similarity predicts marital satisfaction better than
complementation and that neurotic and psychotic characteristics in the
individuals are antithetical. Kelly and Conley (1987) used a longitudinal
design to demonstrate the destructive nature of neuroticism, while Cramer
(1993) found that women who were divorced or separated (but not men) were
higher on N and E than those who continued to be married. While contributing
some additional information, most of the studies that have followed Eysenck
and Wakefield (1981) have been retrograde in the sense that they have not
considered the interaction between husband and wife personality scores and
have thus been unable to address the question of gender-asymmetry
complementation as a principle that qualifies the similarity effect.
 Sex and personality 179

7. GENETIC FACTORS

Eysenck's 1976 book includes an analysis of the responses of twins to his sex
questionnaire, conducted in collaboration with N. G. Martin. With a sample of
153 pairs of male twins and 399 female twin pairs, they reported that in the
case of men, libido was about two thirds determined by additive (dominance
free) genetic factors, while cultural influences were more important for women
(Table 9.5). There was also evidence of a between-family environmental effect
in women, but not in men, suggesting that daughters are more influenced by
family values than sons. Taken together these findings support the hypothesis
that female sexuality is more influenced by social forces than that of men.
With respect to satisfaction, the position was more complicated. There
appeared to be some involvement of genetic factors but environmental influ-
ences were more important for both genders. This makes sense when it is
remembered that sexual fulfillment depends upon the cooperation of other
people (lovers, spouses, etc.), whereas libido is relatively independent of the
behavior of others. In the case of women there was actually a tendency for DZ
twins to be more alike in satisfaction than MZ twins, perhaps implying
competition for scarce resources in the latter group. MZ women probably
attract the same kind of men and move in the same circle of friends, therefore
they would have to compete for attention more than DZ twins. Male twins
perhaps go out and meet people away from home more than female twins, and
hence would not compete to the same extent.
Martin, Eaves, and Eysenck (1977) also found a genetic influence for age of
first intercourse. Since early intercourse was correlated with libido, tough-
mindedness, and extraversion (in both men and women), the genetic effect of
this aspect of sexual behavior could have been mediated by these personality
factors. As regards the environmental component, shared experience (between
family variance) was found to be less important than individual experiences
(within family variance). Again, there was some indication of competition
effects in the female data only.

Table 9.5. Heritability estimates for three aspects of human sexuality—based on samples of 153
males and 339 female twin pairs, respectively (from Eysenck, 1976)

Satisfaction Females .27*

Males .40
Libido Females .45
Males .61
Masculinity versus femininity Females .44
Males .40
*These coefficients can vary between 0 and 1, .50 indicating that half the variance is genetic and
half environmental.
180 Personality

Although they were only able to obtain small samples (14 identical twin pairs
and 14 fraternal), Gosselin and Wilson (1980) replicated Eysenck's finding of a
genetic basis to sex drive and satisfaction and found that paraphilias such as
masochism and transvestitism also showed signs of a genetic component
(higher intraclass correlations for identical twins). Fetishism, however, was a
notable exception, suggesting that imprinting or conditioning processes might
be more involved in this preference.
Also apparently independent of genetic influences is mate selection. Lykken
(1993) studied the characteristics of the spouses of middle aged twins and their
attraction to their twin's spouse, and concluded that mate choice was largely a
matter of random imprinting (whomever happens to be around when Cupid's
arrow strikes). However, divorce does seem to be genetically influenced, as
might be expected from the marital success research described above. McGue
and Lykken (1992) found that the concordance for divorce among MZ twins
was significantly higher than that for DZ twins.

8. SEXUAL CONDITIONING

Eysenck had supposed that the connection between E and sexual behavior was
partly due to individual differences in conditionability. Kantorowitz (1978)
investigated this in an experimental study of the conditioning of sexual arousal,
studying both positive conditioning (showing erotic slides during masturbation
just before orgasm) and deconditioning (showing slides during the resolution
phase just after orgasm). Despite a small sample (8 young men), Kantorowitz
found a significant correlation between E and preorgasmic conditioning of
arousal and a significant connection between introversion and postorgasmic
conditioning. In other words, extraverts were more easily conditioned
positively, but introverts were more easily deconditioned. High N scores had
an effect parallel to that of introversion, but nonsignificantly so.
Kantorowitz's finding might help to explain why extraverts are more sexually
active and adventurous in real life, but does not fit Eysenck's theory that
introverts are generally more conditionable. It is rather more consistent with
Gray's (1973) modification of the Eysenck theory which supposes that
extraverts, being reward-oriented, are more conditionable hi appetitive
contexts, while introverts are more susceptible to punishment and therefore
more conditionable in situations involving guilt and anxiety. Kantorowitz
suggests that introverts might have been more conditionable at both phases
since the experimental procedure is likely to induce some degree of anxiety,
but the high state of sexual arousal prior to ejaculation could have suppressed
that anxiety, thus allowing extraverts to show greater conditioning. After
orgasm, however, anxiety would predominate and there would be nothing to
stop introverts from appearing more conditionable.
 Sex and personality 181

The work of Eysenck and Levey (1972) suggests another possible

explanation of the Kantorowitz finding. Introverts are only found to condition
better than extraverts under conditions of low arousal (e.g., when
reinforcement is weak or partial). When the UCS is very strong extraverts
show superior conditioning. Eysenck uses the concept of protective or
transmarginal inhibition to account for this. The argument is that strong
stimulation may be so aversive to the introvert that a neurological defense
mechanism comes into play which drastically reduces input. Presuming that the
preorgasmic phase is one of exceptionally high arousal, transmarginal
inhibition might reduce the conditionability of introverts to a level below
that of extraverts, whereas introverts would show greater conditioning in the
relative calm of the postorgasm phase.
Whichever interpretation is correct, personality does seem involved in the
conditioning of sexual arousal, and the findings have clinical implications.
Preorgasmic conditioning may be the better means of modifying sexual
preferences with stable, extravert clients, while postorgasmic deconditioning
might produce better results with clients who are emotional introverts.

9. PERSONALITY AND HORMONES

Since both personality and sexual behavior are partly genetic, it is reasonable
to look for some hormonal link between the two. In practice this has proved
difficult for various reasons. (1) Hormones exist in the blood in minute
quantities which are hard to measure accurately. (2) Their level fluctuates from
time to time according to diurnal, seasonal, and menstrual cycles, general
health, age, experiences of success or failure, anticipation of sexual activity, and
so on. (3) Only a small proportion of the hormone is free to act; the rest is
bonded to globulin in such a way as to be inactive. (4) Hormone levels
themselves are not the only critical factor; variations in receptor sensitivity are
equally important and largely independent. (5) Adult levels of circulating
hormone are often less important than those prevailing during prenatal brain
development. (6) Many people take hormone supplements for various reasons
such as contraception, hormone replacement therapy, body building, or
medication. (7) The effects of hormone levels on behavior may be nonlinear,
with very low and high levels producing similar results different from those
observed at intermediate levels (Nyborg, 1994). Given these complications it is
not surprising that sex hormone studies have produced complex and often
contradictory results.
Daitzman (1976) measured androgen and estrogen secretion in student
samples on two separate occasions and correlated the results with a variety of
personality traits, including the EPQ. Since there were only seven women in his
sample, this discussion is restricted to the results for his 76 men. Androgen
182 Personality

secretion was significantly related to N, while correlations with E and P were

positive but nonsignificant. Although a favorable parental attitude to sex was
related to androgen, heterosexual experience was negatively related. Estrogen
levels showed much the same pattern, being correlated with N, while correl-
ations with E and P were positive but small. Correlations with sexual
experience were again negative. When the androgen-estrogen balance was
examined, this was negatively correlated with both masculinity and sexual
experience. These results are almost the opposite of what might have been
expected theoretically.
Somewhat different results were obtained in a later study by Daitzman and
Zuckerman (1980). Within a sample of 40 unmarried male students testo-
sterone was significantly related to E, dominance, heterosexual interest and
experience, and low N (a traditionally male pattern). Estradiol likewise correl-
ated with heterosexual interest and experience, but also with homosexual
experience and various psychopathic and psychotic traits. The authors make no
attempt to explain why these results differ so much from those of Daitzman
(1976).
Persson and Svanborg (1980) studied free (unbound) hormone levels in a
representative sample of 70-year-old men and women, arguing certain advan-
tages in studying older people of restricted age range. Older men, they sug-
gested, would be less likely to be beyond saturation levels of androgens, so
variations would be more likely to have behavioral significance. They found
that a high androgen-estrogen ratio in men was associated with confidence,
dominance, energy, a high frequency of intercourse, and fewer neurotic sym-
ptoms. In the case of women, high estrogen went with low N, affiliative
tendencies (seeking friends and social contacts), and a relatively high freq-
uency of intercourse. These results were summarized by the authors as
suggesting that mental health and positive traits go with a homotypic sex
hormone balance (i.e., a high ratio of male hormones in a man and female
hormones in a woman). There are several possible explanations for this: (1)
having atypical hormones may make adjustment more stressful, leading to
psychiatric symptoms; (2) psychiatric stress may lead to changes in the hor-
mone balance; and (3) mental health, sexuality, and hormone balance may be
jointly affected by the aging process so as to produce the observed correlations.
Many clinical and laboratory studies since then have confirmed that high
androgen levels in men are associated with confidence, competitiveness,
aggression, and even criminality, while androgen deficiency leads to anxiety
and moodiness (Dabbs, Carr, Frady, & Riad, 1995; Dabbs, Hargrove &
Heusel, 1996; Ellis, 1986; Harris, Rushton, Hampson, & Jackson, 1996). The
effect of estrogen on women is less clear, but there is some indication that it
leads to an increase in extraversion and a decrease in neuroticism (Herrmann
& Beach, 1976,1978; Rose, 1972). Estrogen administered to adult men inhibits
libido and aggression.
 Sex and personality 183

It is now widely recognized that prenatal androgens have the capacity to

masculinize a female fetus with ramifications in adult personality. Such females
typically display high levels of energy and aggression, independence and self-
assurance, tomboyish behavior and reduced interest in the maternal role (D.
M. Quadagno, Briscoe, & J. S. Quadagno, 1977; Reinish, 1977). Also relevant
is the finding of Wilson (1984) that deeper-voiced opera singers (between- and
within-gender) are less emotional and more sexually predatory than their
higher-voiced counterparts, associations that appear to be mediated by sex
hormone effects during development of the brain and vocal structures.
The field of the biochemical basis of sexuality and personality is fast
developing but current indications are that androgens are involved in libido,
masculinity, and psychoticism (and perhaps certain aspects of extraversion as
well), while estrogen is in some way connected with affiliation, empathy, and
sexual receptivity in women. Deficiencies in either seem to be connected with
negative mood states such and anxiety and depression, and failure to perform
the appropriate sex role, especially when it is the "homotypic" hormone that is
lacking. The relative contributions of prenatal secretion, circulating levels, and
receptor sensitivity are still not fully understood, and Nyborg (1994) makes a
powerful case for examining personality and other attributes in relation to
various levels of sex hormones (androtypes among males and estrotypes among
females) since hormone effects often appear to be nonlinear. In particular, he
suggests that sociability is related to high androgen levels in men and high
estrogen levels in women (i.e. the homotypic hormones).

10. CONCLUSION

Hans Eysenck has been a major contributor to research on the way in which
personality influences sexual attitudes and behavior and marital choice and
satisfaction. Generally speaking, the extrapolation of personality into the field
of sexual behavior follows predictable lines. Extraverts pursue an active,
sociable, variable, and pleasure-oriented sex life, while introverts are relatively
quiet, controlled, private, and discriminating. High N people show anxiety and
conflict in sexual matters and so are more prone to dysfunctions such as
orgasm difficulty, erection failure, and dissatisfaction, whereas low N people
are more stable, free of performance problems and contented with their sex
lives. High P scorers extend their unconventional and impersonal tendencies
into their sex lives, seeking sensation and gravitating towards impersonal,
variant preferences; low P scorers (rather like high L scorers and con-
servatives) are more conventional, considerate, and seeking of intimacy.
Differences between men and women largely parallel the differences between
high and low P scorers, suggesting that androgens (whether prenatal or
contemporary) are involved with both masculinity and psychoticism.
184 Personality

Personality also influences the kind of deviant sexual behaviors that are
adopted by some people, particularly men. Rapists appear high on libido, but
lacking in the social inhibitions that prevent most men from "stealing" sex.
They thus tend to be high on P and E (impulsive and lacking in socialization).
Sexual variations characterized by impersonal outlets (e.g., fetishism, maso-
chism, and transvestitism), and also pedophilia, are more often adopted by
men who are afraid of social contact (thus usually introverted and high on N).
These deviant forms of sexual expression are largely male, probably because
females have a less predatory (targeting) form of sex drive. Women's diffi-
culties are more likely to be in forming satisfying long-term relationships
(attractiveness and social skills) or in obtaining sufficient arousal and orgasm.
Women who fail to achieve satisfaction and have orgasm difficulties are likely
to be high on N and perhaps tending towards introversion and low P. The most
common forms of sexual dysfunction in men (e.g., premature or retarded
ejaculation and erectile problems) are also associated with high N and to a
lesser extent introversion. Clinicians who acquaint themselves with these
findings will be better placed to understand the sexual problems of their
patients.
A major social lesson to be learned from these findings is that no rigid moral
prescription can easily accommodate the striking individual differences that
occur in people's sexual needs and preferences. Puritanism may suit introverts,
older people, and a high proportion of women, but is an impossible code for
others to sustain. Permissiveness suits extraverts, high P individuals, young
people, and many men, but its public manifestation easily becomes offensive to
others. There is also the risk that permissiveness can (paradoxically) become a
kind of repressive orthodoxy, pushing naturally reserved people towards
activities they find distasteful.
Eysenck's work on personality in relation to marital success is an important
step forward in resolving the long-standing dispute between similarity and
complementation theories of partner choice. It shows why correctional studies
have consistently supported similarity theory while at the same time explaining
why complementation theory is so enduring (it contains and element of truth in
that gender typical differences are best matched within individual partner-
ships). Against a general background of homogamy there are certain masculine
and feminine traits that complement each other. Eysenck's research in also
valuable in reminding us that certain traits brought to the partnership by each
individual predict marital unhappiness regardless of the relationship chemistry
and that marriage serves different needs for men and women. Eysenck's work
is more sophisticated than most preceding studies in viewing marital success
from the perspectives of husband and wife separately.
 Sex and personality 185

REFERENCES

Apt, C., & Hurlbert, D. F. (1994). The sexual attitudes, behaviour and relationships of
women with histrionic personality disorder. Journal of Sex and Marital Therapy, 20,
125-133.
Bentler, P. M., & Newcomb, M. D. (1979). Longitudinal study of marital success and
failure. In M. Cook & G. D. Wilson (Eds.), Love and attraction. Oxford: Pergamon
Press.
Bentler, P. M., & Peeler, W. H. (1979). Models of female orgasm. Archives of Sexual
Behaviour, 8, 405-423.
Catlin, N., Croake, J. W., & Keller, J. F. (1976). MMPI profiles of cohabiting college
couples. Psychological Reports, 38, 407-410.
Cramer, D. (1993). Personality and marital dissolution. Personality and Individual
Differences, 14, 605-607.
Dabbs, J. M., Carr, T. S., Frady, R. L., & Riad, J. K. (1995). Testosterone, crime and
misbehaviour among 692 male prison inmates. Personality and Individual Differences,
18, 627-633.
Dabbs, J. M., Hargrove, M. F., & Heusel, C. (1996). Testosterone differences among
college fraternaties: well-behaved vs rambunctious. Personality and Individual
Differences, 20, 157-161.
Daitzman, R. J. (1976). Personality correlates of androgens and oestrogens.
Unpublished doctoral dissertation, University of Delaware.
Daitzman, R. J., & Zuckerman, M. (1980). Disinhibitory sensation-seeking, personality
and gonadal hormones. Personality and Individual Differences, 1, 103-110.
Ellis, L. (1986). Evidence of neurandrogenic etiology of sex roles from a combined
analysis of human, non-human and non-primate mammalian studies. Personality and
Individual Differences, 7, 519-552.
Eisinger, A. J., Huntsman, R. G., Lord, J., Merry, J., Polani, P., Tanner, J. M.,
Whitehouse, H., & Griffiths, P. D. (1972). Female homosexuality. Nature, 106, 238.
Evans, R. B. (1970). Sixteen personality factor questionnaire scores of homosexual men.
Journal of Consulting and Clinical Psychology, 34, 212-215.
Eysenck, H. J. (1976). Sex and personality. London: Open Books.
Eysenck, H. J. (1979). The structure and measurement of intelligence. Berlin: Springer.
Eysenck, H. J. (1980). Personality, marital satisfaction and divorce. Psychological
Reports, 47, 1235-1238.
Eysenck, H. J., & Levey, A. (1972). Conditioning, introversion-extraversion and the
strength of the nervous system. In V. D. Neblitsin & J. A. Gray (Eds.), Biological
bases of individual behaviour. New York: Academic Press.
Eysenck, H. J., & Wakefield, J. A. (1981). Psychological factors as predictors of marital
satisfaction. Advances in Behaviour Research and Therapy, 3, 151-192.
Eysenck, H. J., & Wilson, G. D. (1979). The psychology of sex. London: Dent.
Farkas, G. M., Sine, L. F., & Evans, I. M. (1979). The effects of distraction,
performance demand, stimulus explicitness and personality on objective and
subjective measures of male sexual arousal. Behaviour Research and Therapy, 17,
25-32.
Fisher, S. (1973). Understanding the female orgasm. Harmondsworth: Penguin.
Giese, H., & Schmidt, A. (1968) Studenten Sexualitdt. Hamburg: Rohwolt.
186 Personality

Gosselin, C. C, & Eysenck, S. B. G. (1980). The transvestite "double-image": a

preliminary report. Personality and Individual Differences, 1, 172-173.
Gosselin, C. C., & Wilson, G. D. (1980). Sexual variations: Fetishism, sadomasochism
and transvestism. New York: Simon & Schuster.
Gosselin, C. C., Wilson, G. D., & Barrett, P. T. (1991). The personality and sexual
preferences of sadomasochistic women. Personality and Individual Differences, 12,11-
15.
Gray, J. A. (1973). Causal theories of personality and how to test them. In J. R. Royce
(Ed.), Multivariate analysis and psychological theory. New York: Academic Press.
Griffith, M., & Walker, C. E. (1975). Menstrual cycle phases and personality variables
as related to response to erotic stimuli. Archives of Sexual Behaviour, 4, 599-603.
Harris, J. A., Rushton, J. P., Hampson, E., & Jackson, D. N. (1996). Salivary
testosterone and self-report aggressive and pro-social personality characteristics in
men and women. Aggressive Behaviour, 22, 321-331.
Herrmann, W. M., & Beach, R. C. (1976). Psychotropic effects of androgens: a review
of clinical observations and new human experimental findings. Pharmacopsychiatry, 9,
205-219.
Herrmann, W. M., & Beach, R. C. (1978). The psychotropic properties of oestrogens.
Pharmacopsychiatry, 11, 164-176.
Husted, J. R., & Edwards, A. E. (1976). Personality correlates of male sexual arousal
and behaviour. Archives of Sexual Behaviour, 7, 149-156.
Insel, P. M. (1971). Family similarities in personality, intelligence and social attitudes.
Unpublished doctoral dissertation, University of London.
Joe, V. C., Brown, C. R., & Jones, R. (1976). Conservatism as a determinant of sexual
experience. Journal of Personality Assessment, 40, 516-521.
Kane, F. J., Lipton, M. A., & Ewing, J. A. (1969). Hormonal influences in female sexual
response. Archives of General Psychiatry, 20, 202-209.
Kantorowitz, D. A. (1978). Personality and conditioning of tumescence and detumes-
cence. Behaviour Research and Therapy, 6, 117-123.
Kelly, E. L., & Conley, J. J. (1987) Personality and compatibility: A prospective analysis
of marital stability and marital satisfaction. Journal of Personality and Social
Psychology, 52, 27-40.
Kennedy, P., Doherty, N., & Barnes, J. (1995). Primary vaginismus: A psychometric
study of both partners. Sexual and Marital Therapy, 10, 9-22.
Kirton M. (1977). Relatedness of married couples' scores on Adosno-type tests.
Psychological Reports, 40, 1013-1014.
Lucas, A., & Peterson, G. W. (1991). MMPI scales as predictors of marital locus of
control and marital decision-making satisfaction in women. Personality and Individual
Differences, 12, 1219-1225.
Lykken, D. T. (1993). Is human mate selection adventitious or the result of lawful
choice? A twin study of mate selection. Journal of Personality and Social Psychology,
65, 56-68.
Lynn, R., & Martin, T. (in press). Gender differences in extraversion, neuroticism, and
psychoticism in 37 nations. Journal of Social Psychology.
McGue, M., & Lykken, D. T. (1992). Genetic influence on risk of divorce. Psychological
Science, 3, 368-373.
 Sex and personality 187

Martin, N. G., Eaves, L. J., & Eysenck, H. J. (1977). Genetical, environmental and
personality factors influencing the age of first intercourse in twins. Journal of
Biosocial Science, 9, 91-97.
Medicus, G., & Hopf, S. (1990). The phylogeny of male/female differences in sexual
behaviour. In J. R. Feierman (Ed.), Paedophilia: Biosocial dimensions. New York:
Springer Verlag.
Munjack, D. J., Kanno, P. H., & Oziel, L. J. (1978). Ejaculatory disorders: some
psychometric data. Psychological Reports, 43, 783-789.
Nias, D. K. B. (1977). Husband-wife similarities. Social Science, 52, 206-211.
Nyborg, H. (1994). Hormones, sex, and society: The science ofphysicology. Westport, CT:
Praeger.
O'Sullivan, D. M., Zuckerman, M., & Kraft, M. (1996). The personality of prostitutes.
Personality and Individual Differences, 21, 445-448.
Persson, G., & Svanberg, A. (1980). Personality, mental symptoms and sexuality in
relation to androgen and oestrogen activity in plasma in a 70-year old population.
Unpublished paper, Gothenburg, Sweden.
Quadagno, D. M., Briscoe, R., & Quadagno, J. S. (1977). Effect of perinatal gonadal
hormones on selected non-sexual behaviour patterns: a critical assessment of the
nonhuman and human literature. Psychological Bulletin, 84, 62-80.
Reinisch, J. M. (1977). Prenatal exposure of human foetuses to synthetic progestin and
oestrogen affects personality. Nature, 266, 561-562.
Richard, L. S., Wakefield, J. A., & Lewak, R. (1990). Similarity of personality variables
as predictors of marital satisfaction: A Minnesota Multiphasic Personality Inventory
(MMPI) item analysis. Personality and Individual Differences, 11, 39-43.
Rose, R. M. (1972). The psychological effects of androgens and oestrogens: a review. In
R. I. Shader (Ed.), Psychiatric complications of medical drugs. New York: Raven
Press.
Rushton, J. P., Russell, R. J. H., & Wells, P. A. (1984). Genetic similarity theory:
Beyond kin selection. Behaviour Genetics, 14, 179-193.
Russell, R. J. H., & Wells, P. A. (1991). Personality similarity and quality of marriage.
Personality and Individual Differences, 12, 407-412.
Russell, R. J. H., & Wells, P. A. (1994). Predictors of happiness in married couples.
Personality and Individual Differences, 17, 313-321.
Schenk, J., & Pfrang, H. (1986). Extraversion, neuroticism and sexual behaviour:
Interrelationships in a sample of young men. Archives of Sexual Behaviour, 15, 449-
455.
Schmidt, G., Sigusch, V., & Meyberg, U. (1969). Psychosexual stimulation in men:
emotional reactions, changes of sex behaviour and measures of conservative
attitudes. Journal of Sex Research, 5, 199-217.
Schofield, M. (1968). The sexual behaviour of young people. Harmondsworth: Penguin.
Shope, D. F. (1968). The orgastic responsiveness of selected college females. Journal of
Sex Research, 4, 204-219.
Thomas, D. R. (1975). Conservatism and premarital sexual experience. British Journal of
Social and Clinical Psychology, 14, 195-196.
Thomas, D. R., Shea, J. D., & Rigby, R. G. (1971). Conservatism and response to sexual
humour. British Journal of Social and Clinical Psychology, 10, 185-186.
188 Personality

Thome, F. C, Haupt, T. D., & Allen, R. M. (1966). Objective studies of adult male
sexuality utilising the Sex Inventory. Journal of Clinical Psychology Monograph
Supplement, 21, October.
Wilson, G. D. (1978). The secrets of sexual fantasy. London: Dent.
Wilson, G. D. (1984). The personality of opera singers. Personality and Individual
Differences, 5, 195-201.
Wilson, G. D. (1987). Biological and social aspects of female sexual functioning. In E.
Karas (Ed.) Current issues in clinical psychology (Vol. 3). New York: Plenum.
Wilson, G. D. (1989). The great sex divide. London: Peter Owen.
Wilson, G. D., & Cox, D. N. (1983). Personality of paedophile club members.
Personality and Individual Differences, 4, 323-329.
Wilson, G. D., & Fulford, K. W. M. (1979). Sexual behaviour, personality and hormonal
characteristics of heterosexual, homosexual and bisexual men. In M. Cook & G. D.
Wilson (Eds.), Love and attraction. Oxford: Pergamon.
Wilson, G. D., & Gosselin, C. C. (1980). Personality characteristics of fetishists,
transvestites and sadomasochists. Personality and Individual Differences, 1, 289-295.
Wilson, G. D., & MacLean, A. (1974). Personality, attitudes and humour preferences of
prisoners and controls. Psychological Reports, 34, 847-854.
Zaleski, Z., & Galkowska, M. (1978). Neuroticism and marital satisfaction. Behaviour
Research and Therapy, 16, 285-286.
Zuckerman, M. (1974). The sensation-seeking motive. In B. Maher (Ed.), Progress in
experimental personality research, 7, New York: Academic Press.
Zuckerman, M., Tushup, R., & Finner, S. (1976). Sexual attitudes and experiences:
attitudes and personality correlates and changes produced by a course in sexuality.
Journal of Consulting and Clinical Psychology, 44, 7-19.
Chapter 10
Extraversion and impulsivity: The lost
dimension?*

W. Revelle

... quick intelligence, memory, sagacity, cleverness, and similar qualities, do not often
grow together, and ... persons who possess them and are at the same time high-
spirited and magnanimous are not so constituted by nature as to live in an orderly
and peaceful and settled manner; they are driven any way by their impulses, and all
solid principle goes out of them.
... On the other hand, those stable and steadfast and, it seems, more trustworthy
natures, which in a battle are impregnable to fear and immovable, are equally
immovable when there is anything to be learned; they are always in a torpid state, and
are apt to yawn and go to sleep over any intellectual toil. (Plato, The Republic, Book 6
503c from Benjamin Jowett 4th ed.)

1. INTRODUCTION

Over 2000 years after Plato described a dimension of impulsivity, psychologists

are still concerned with those who are unable to live in an orderly and peaceful
manner. Impulsive children and adults are carefree, long for excitement, act
rapidly and without thinking, and respond to dares and challenges. Compared
to the less impulsive, they are more likely to be found sky diving or hang
gliding, to have automobile accidents and traffic violations, to be arrested, to
commit violent suicide, and to perform better under high time stress
conditions. In childhood, impulsivity is linked to difficulties in sustained
attention and is a core feature of the diagnostic category of Attention Deficit
Disorder. In adulthood, impulsivity is linked to behavior difficulties and to
psychopathy.
*Preparation of this chapter was supported in part by contract MDA903-93-K-0008 from the U.S.
Army Research Institute to William Revelle and Kristen Anderson. The views, opinions, and
findings contained in this chapter are those of the author and should not be construed as an official
Department of the Army position, policy, or decision, unless so designated by other official
documentation.
190 Personality

Impulsivity has had a varied role in the study of personality and

temperament and an even more varied role in the work of Hans Eysenck.
While impulsivity was at one point a core feature of extraversion (Eysenck,
1967; H. J. Eysenck & S. B. G. Eysenck, 1967; S. B. G. Eysenck & H. J.
Eysenck, 1963, 1969) that was said to be married in a shotgun wedding with
Sociability (Guilford, 1975), it is now seen by some as a component of
Psychoticism (Eysenck, 1990, 1991, 1992; H. J. Eysenck & M. W. Eysenck,
1985) and by others as a central component of uncontrolled stimulation
seeking and psychopathic behavior (P-Imp-USS, Zuckerman, 1994). Consider-
ed a facet of emotionality (Costa & McCrae, 1992) or nonconscientiousness
(Digman, 1994), impulsivity has had a varied life in its identification in three-,
four-, or five-dimensional personality space. At the same time as it has suffered
from an identity crisis in terms of measurement (Rocklin & Revelle, 1981),
impulsivity has been identified as a central feature in arousal based theories of
cognitive performance (Anderson & Revelle, 1994; Humphreys & Revelle,
1984; Revelle, 1989; Revelle, Humphreys, Simon, & Gilliland, 1980) that has a
strong biological basis (Schalling & Asberg; 1985; Zuckerman, 1991). This
chapter reviews a small part of the extensive literature on impulsivity and
extraversion, and discusses the vital research contribution that Hans Eysenck
and his colleagues have made to understanding this important personality trait.

2. IMPULSIVITY AND THE STRUCTURE OF PERSONALITY

2.1 Introduction
In the first three-dimensional model of temperament, Heymans related
impulsivity to a bias towards the primary effects of stimulation versus a bias
towards secondary or reflective processing (Heymans, 1929 as cited by Van der
Werff, 1985; Van der Werff & Verster, 1987). Considering secondary
functioning, in combination with two other dimensions, emotionality and
activity, allowed Heymans to go beyond the personality types of Galen and to
introduce dimensional thinking into personality research. In later reanalyzes of
Heymans' data, ratings of impulsivity had high loadings on the "secondary
function" factor and were associated with being lively and busy, demonstrative,
violent, but not calm, quiet, or thoughtful (Van der Werff & Verster, 1987).
In his Explorations of Personality, Murray considered "Impulsion" as "the
tendency to respond ... quickly and without reflection" and as characterizing
someone who "is usually restless, quick to move, quick to make up his mind,
quick to voice his opinion. He often says the first thing that comes into his
head; and does not always consider the future consequences of his conduct"
(Murray, 1938, p. 205).
 Extroversion and impulsivity: The lost dimension? 191

2.2 Impulsivity and Extraversion—the early years

Although impulsivity does not appear in the index of The Dimensions of
Personality (Eysenck, 1947), within nine years the basic self-report measure of
extraversion was a mixture of sociability and impulsivity (Eysenck, 1956). Two
years later, impulsivity represented three of the six items in a short measure of
extraversion (Eysenck, 1958; see Table 10.1). It is interesting to note that
sociability, which would come to play such a dominant part of the Extraversion
construct, was not considered an important component of extraversion in 1947
and was at most half of Extraversion in 1958.
By 1960, the Maudsley Personality Inventory (Eysenck, 1959) extraversion
scale was criticized as being factorially complex and as representing a mixture
of sociability and impulsivity (Carrigan, 1960). To the Eysencks this mixture of
impulsivity and sociability represented the dual nature (S. B. G. Eysenck & H.
J. Eysenck, 1963, 1969) of a unitary dimension (H. J. Eysenck & S. B. G.
Eysenck, 1967, 1969). The Eysencks as well as Sparrow and Ross (1964)

Table 10.1. Representative Impulsivity items taken from Murray (1938), Eysenck (1958) short
form, the "dual nature" (S. B. G. Eysenck & H. J. Eysenck, 1963), and the EPI (H. J. Eysenck &
S. B. G. Eysenck, 1964). Reversed items are marked with (R)

Item # Item Source

I often act on the spur of the moment without stopping to think Murray
I waste no time in asking for what I want Murray
I often act impulsively just to blow off steam Murray
I usually make a plan before I start to do something (R) Murray
I do most things slowly and deliberately (R) Murray
I am slow to decide on a course of action (R) Murray
B Do you prefer action to planning for action? Short Form
D Are you happiest when you get involved in some project that Short Form
calls for rapid action?
H Are you inclined to be quick and sure in your actions? Short Form
14 Do you often act on the spur of the moment without stopping Dual Nature
to think?
22 Are you inclined to stop and think things over before acting? Dual Nature
35 Would you describe yourself as an easy going person not Dual Nature
concerned to be precise?
36 Do you tend towards a rather reckless optimism? Dual Nature
40 Are you given to acting on impulses of the moment which Dual Nature
later land you in difficulties?
50 Do you prefer action to planning for action? Dual Nature
1 Do you often long for excitement? EPI
3 Are you usually carefree? EPI
5 Do you stop and think things over before doing anything? (R) EPI
8 Do you generally do and say things quickly without stopping EPI
to think?
10 Would you do almost anything for a dare? EPI
13 Do you often do things on the spur of the moment? EPI
22 When people shout at you, do you shout back? EPI
39 Do you like doing things in which you have to act quickly? EPI
41 Are you slow and unhurried in the way you move? (R) EPI
192 Personality

showed that Extraversion items form two correlated factors as would be

predicted from the hierarchical formulation of Extraversion (Eysenck, 1967).
(It is interesting to compare the hierarchical model from 1947 with that of 1967
and 1969. In 1947 Introversion at the "Type" level was made up of the "Traits"
of Persistence, Rigidity, Autonomic Imbalance, Accuracy, and Irritability. In
1967 and 1969 a similar appearing figure showed Extraversion at the Type level
to be composed of Sociability, Impulsiveness, Activity, Liveliness, and Excita-
bility. The 1967 figure is said to be "reprinted with slight changes" from the
1947 text. In fact, except for the structural characteristics of a hierarchy ranging
from specific responses to habitual responses to traits to types, there seems to
be no overlap between the two conceptions of Extraversion. This subtly
changing nature of extraversion and the place of impulsivity within Eysenck's
theory would continue to be a question for the next 30 years.)
In the following years the Maudsley Personality Inventory was modified to
improve the factor structure of the Extraversion scale and to increase the
independence of E and N. An early revision, the Eysenck Personality Inventory
(EPI; H. J. Eysenck & S. B. G. Eysenck, 1964) had a 24-item Extraversion scale
made up predominantly of Sociability and Impulsivity items (Table 10.1). With
this combination of the subdomains, E was almost orthogonal to N. Over the
next 10 years the EPI was the operational definition of Extraversion and was
the basis for a great deal of genetic, physiological, and cognitive research.
Some of this work was reviewed in The Biological Basis of Personality
(Eysenck, 1967), an impressive summary of the behavioral and biological
correlates of Extraversion and Neuroticism that provided a road map for the
next three decades of research on Extraversion. In addition to reviewing the
many physiological correlates of Extraversion and Neuroticism, Eysenck laid
out the fundamental hypotheses about the relationship between Extraversion
and arousal that would be the core of experimental and physiological research
on extraversion. What is interesting in retrospect is that the book did not,
however, make any distinction between impulsivity and sociability, and in fact
rejected as implausible any consideration of a rotation from the basic N and E
dimensions.

2.3 Impulsivity and the dual nature of extraversion

As experimental research focused on the behavioral, cognitive, and physiolo-
gical correlates of Extraversion, psychometric research focused on difficulties
in its measurement. The EPI-E scale was criticized by Guilford (1975) as
representing a "shotgun marriage" of sociability and impulsivity, a position
that Eysenck (1977) strongly rejected. Although the centroid of impulsivity
items was about 60 degrees away from that of a set of sociability items,
Guilford (1977) argued that it was possible to recover pure and orthogonal
measures of his R factor (Restraint vs. Rhathymia) and S (Sociability) factors.
 Extroversion and impulsivity: The lost dimension? 193

To Guilford, R was the true measure of Extraversion. (Using Guilford's

rotations of the Eysenck items, and referring to the Eysenck and Eysenck 1969
analysis, the highest loading items on the R scale are 22, 40, 14, 50 and 35,
Table 10.1). Although it is clear that the items used to measure E range across
90 degrees, it is also the case that most of the items were within 30 degrees of
the central Extraversion factor (H. J. Eysenck & S. B. G. Eysenck, 1969; S. B. G.
Eysenck & H. J. Eysenck, 1969).
Until about 1975, Impulsivity (Imp) and Sociability (Soc) were seen as the
defining components of Extraversion. This was made particularly explicit in a
genetic analysis of the heritabilities of the two subscales from the EPI as well as
the combined Extraversion factor (Eaves & Eysenck, 1975). While both Imp
and Soc had roughly equal heritabilities (.6 when correcting for unreliability),
and there was a reliable genetic component to their intercorrelation, there was
a larger environmental component to the phenotypic correlation. This result
led Eaves and Eysenck to suggest that the unitary nature of E at the phenotypic
level was more strongly due to environmental rather than genetic factors.
A second influential model of the biological basis of personality that
emphasized impulsivity rather than extraversion but that stayed in the same
two-dimensional space was proposed by Gray (1972; see chapters 1-3 in this
volume) who suggested a 45 degree rotation of the E/N axes to highlight
anxiety (thought to be high N, low E) and impulsivity (high N, high E). Anxiety
was hypothesized to represent a Behavioral Inhibition System (BIS) while
impulsivity was hypothesized to represent the activation of a Behavioral
Activation System (BAS). This model and its subsequent modifications and
revisions (Gray, 1981, 1991, 1994) has become one of the standard biological
models of personality (Revelle, 1995). Eysenck (1987) criticized the resulting
emphasis upon the primary traits of anxiety and impulsivity (e.g., Barratt, 1987;
Fowles, 1987; Revelle, 1987) and recommended focusing on the higher order
dimensions of E and N.
Anticipating Gray's causal rotation of the E/N axes to represent Impulsivity
and Anxiety, Kassenbaum, Couch, and Slater (1959) showed that a two-
dimensional solution to the MMPI could be thought of in terms of I-E and S-
N or rotated 45 degrees to emphasize impulsivity (vs. intellectual control) and
social participation (vs. social withdrawal). This rotation emphasized the
negative emotional consequences of high impulsivity.

2.4 Impulsivity and the P-E-N model

Further psychometric refinements of the EPI and the introduction of a
Psychoticism scale led to the Eysenck Personality Questionnaire (EPQ; H. J.
Eysenck & S. B. G. Eysenck, 1975; see chapter 6 in this volume) which was to
measure the three factors of the Psychoticism-Extraversion-Neuroticism
model. Although the EPQ was said to provide parallel scales of E and N
194 Personality

with the EPI, in fact the E scale had a markedly different item content
(Rocklin & Revelle, 1981). Seven of the nine impulsivity items in the EPI-E
scale vanished, one (no. 39) stayed in the revised E scale, and one (no. 5)
appeared on the new Psychoticism scale (Campbell & Reynolds, 1984). The
situation did not change with the subsequent revision to the P scale and the
release of the EPQ-R (S. B. G. Eysenck, H. J.Eysenck, & Barratt, 1985; Roger
& Morris, 1991)
Contrary to following Guilford's advice to emphasize the unrestrained
(impulsive) part of E, it seems that the Eysencks had decided to focus on the
Sociability part of E and to claim that impulsivity was a component of P. A
comparison of the EPI, EPQ, and Guilford scales reported that while
Extraversion as measured by the EPI correlates .58 with Guilford's Sociability
(S) scale, EPQ-E correlates .81 with S. In opposite fashion, the EPI-E
correlation with Guilford's Restraint-Rhathymia (R) scale of .61 drops to .41
for the EPQ-E scale, "...what Eysenck is currently referring to as Extraversion
is quite similar to what Guilford calls Social Activity and has only a weak
relationship with what Guilford calls Introversion-Extraversion" (Campbell &
Reynolds, 1984, p 316; see also Amelang & Ullwer, 1991). The importance of
this change in the salience of the traits of sociability and impulsivity in the
meaning of the "type" of extraversion may be seen in their differential pattern
of correlations with preferences for cooperative and competitive activities
(Wolfe & Kasmer, 1988).
In addition to the development of measures of PEN, efforts were made to
develop supplementary items to measure Impulsivity. Multiple revisions of an
expanded impulsivity scale were developed, the best known two of which were
the I5 (S. B. G. Eysenck & H. J. Eysenck, 1978) and the I7 (S. B. G. Eysenck et
al. 1985). Factor analyses of these scales showed that impulsivity was factorially
complex with usually about four subfactors. Narrow impulsiveness,
nonplanning, liveliness, and risk-taking scales were found to be reliable and
moderately correlated. Just as each of these four subscales have different
patterns of correlations with PEN (mainly positively correlated with P and E
with narrow Imp also correlating with N), so did they have different
relationships with performance measures. The move towards psychometric
refinement did not necessarily lead to higher predictive validities. In a review
of the relationship of impulsivity to conditioning, Frcka and Martin (1987)
conclude that the narrow impulsivity items from the EPI show a more
consistent pattern of interactions with stimulus patterns than the revised scales
found on the I5 or I7.

2.5 Alternative measures of impulsivity

An unfortunate tendency in personality research is to develop new scales to
measure old constructs. One reason for this is psychometric refinement,
 Extroversion and impulsivity: The lost dimension? 195

another is for more precise theoretical specification of constructs. Perhaps the

largest is to stamp one's individuality (and scale?) on one's research. Such
multiplicity of scales can lead to confusion as identical constructs are assessed
with different scales or as different constructs are measured by scales with
similar labels. Just as personality scales have proliferated in other areas, so
have they in impulsivity research.
Many measures of impulsivity, from Murray's original explorations of
impulsion (1938), to Guilford's dimensions of temperament (Guilford &
Zimmerman, 1949), to Eysenck (1956) to Zuckerman (1994), ask for variations
on the basic self-descriptive items "are you an impulsive person" and "do you
do and say things without stopping to think" (Table 10.2). Variations on these
items emphasize motoric, cognitive, and affective impulsivity as part of a
general action orientation (Barratt, 1987; Barratt & Patton, 1983). The Barratt
scales are highly correlated with the impulsivity scores from the I7 and the
structure of the pooled items suggests dimensions of rapid decision making
and lack of foresight (Luengo, Carrillo-de-la-Pefia, & Otero, 1991). Rating
scales for children and self-report inventories for adults were developed to
assess the development of four temperaments, including impulsivity (Buss &
Plomin, 1975). The impulsivity scale of the EAST (Emotionality, Activity,
Sociability, Impulsivity) had four components, reflecting differences in
inhibitory control, decision time, sensation seeking, and persistence. The
Impulsivity scale of the Karolinska Scales of Personality included items
emphasizing nonplanning, rapid decision making, and carefree behavior taken
from Guilford and Barratt (Schalling & Asberg, 1985; Schalling, Edman &
Asberg, 1983). As would be expected given the similar source of items, the
KSP-Imp scale is highly correlated with the EPI-imp scale. A scale composed
of protypical acts of impulsivity correlates with EPQ-E and N as well as other
standard measures of impulsivity (Romero et al, 1994).
Behaviorally, impulsivity as contrasted to reflection is said to result in rapid
but inaccurate performance on a visual perception task (Kagan, 1966).
However, scores on the Matching Familiar Figures Test show low correlations
with most self-report measures of impulsivity (Gerbing, Ahadi, & Patton, 1987;
Helmers, Young & Pihl, 1995). Rather than reflecting an overall difference in
speed of response, impulsivity as indexed by the combination of high
Neuroticism and high Extraversion (i.e., impulsivity as specified by Gray,
1972) leads to an inability to change the speed of response when told to draw a
figure as slowly as possible (Bachorowski & Newman, 1985, 1990). That is,
impulsivity is an inability to inhibit responding rather than just a fast rate of
responding.
When working on simple cognitive tasks, faster performance usually results
in a higher error rate. The appropriate rate of performance represents a
balance between the rewards for the number of problems that are correct and
the penalty for incorrect answers (Revelle, 1986). More impulsive subjects are
196 Personality

Table 10.2. Representative items from different components of impulsivity

Author Component Example items

Barratt (1987) Motoric I do things without thinking
Cognitive I make up my mind quickly
Nonplanning I plan trips well ahead of time (R)
Inhibitory control I have trouble controlling my impulses
Decision time I often act on the spur of the moment
Buss and Plomin Sensation seeking I generally seek new and exciting experiences and
(1975) sensations
Persistence Once I get going on something I hate to stop (R)
Dickman (1990) Functional I don't like to do things quickly, even when I am doing
impulsivity something that is not very difficult (R)
Dysfunctional Often, I don't spend enough time thinking over a situation
impulsivity before I act
Narrow Do you often buy things on impulse? Do you generally do
and say things without stopping to think? Are you an
impulsive person? Do you often do things on the spur of
the moment? Do you get extremely impatient if you are
kept waiting by someone who is late?
Risk taking Do you enjoy taking risks? Would you do almost anything
for a dare? Do you often long for excitement?
S. B. G. Eysenck Nonplanning Do you like planning things carefully well ahead of time?
and H. J. Eysenck (R). When buying things, do you usually bother about the
(1977) guarantee? When you go on a trip, do you like to plan
route and timetables carefully?
Liveliness Do you usually make up your mind quickly? Are you slow
and unhurried in the way you move (R); Do you prefer to
"sleep on it" before making decisions? Can you put your
thoughts into words quickly?
Spontaneous I act on impulse
Gerbing, Ahadi, Not persistent You have a habit of starting things and then losing interest
and Patton (1987) in them
Carefree I am happy-go-lucky
Parker, Bagby, and Cautious versus I think before doing something
Webster (1993) spontaneous
Methodical versus I am very serious minded (R)
disorganized
Schalling & Asberg Nonplanning Do you more often make up your mind quickly than
(1985) working out a decision slowly and carefully?
Rapid decision When I have to make a decision, I "sleep on it" before I
making decide (R)
Carefreeness I take life easy

more likely to adopt a style of faster responses and a higher error rate than are
low impulsives (Dickman & Meyer, 1988; Rawlings, 1984). Moreover, when
considering the relationship between speed of processing and the resulting
error rate, two components of impulsivity can be identified (Dickman, 1990).
Functional impulsivity is associated with rapid responses when they are
 Extraversion and impulsivity: The lost dimension? 197

appropriate, while dysfunctional impulsivity seems to be an inability to adapt to

an optimal response rate (Brunas-Wagstaff, Bergquist, & Wagstaff, 1994;
Dickman, 1990). Functional impulsivity is positively related to EPQ-E and P
and negatively to EPQ-N, while dysfunctional impulsivity is more related to E
and P but not to N (Brunas-Wagstaff, Bergquist, Richardson, & Connor,
1995). Dysfunctional impulsivity is more related to EPI-Imp and the Barratt
Impulsivity Scale than functional impulsivity (Dickman, 1990).
The location of impulsivity in the five-dimensional models known variously as
the "Big 5" (Goldberg, 1990; John, 1990), the "Five Factor Model" (Costa &
McCrae, 1992), and the "alternate Big 5" (Zuckerman, Kuhlman, Joireman,
Teta, & Kraft, 1993) varies by model specification and theorist. Although showi-
ng that EPI-Soc and EPI-Imp were strong markers for extraversion in the
extended NEO personality inventory (McCrae & Costa, 1985), later development
of the NEO located impulsivity as a facet of neuroticism that was equally
correlated with E and N (Costa & McCrae, 1992). Other markers for impulsivity
were correlated with (non)Conscientiousness, Extraversion, and Openness
(Costa & McCrae). Within the lexical tradition of the "Big 5" impulsivity is
seen as a mixture of (non)Conscientiousness and (non)Emotional stability
(Hofstee, de Raad, & Goldberg, 1992). Impulsivity in adolescence is seen as
representing (non)Conscientiousness and may be assessed using ratings adapted
from the Childs California Q Sort (John et al., 1994).
At the scale level, Parker, Bagby, and Webster (1993) report one-, two- and
three-dimensional solutions to impulsivity measures from the Personality
Research Form (PRF: Jackson, 1984), the Multidimensional Personality
Questionnaire (MPQ: Tellegen, 1982, 1985), and the Guilford-Zimmerman
Temperament Survey (GZTS: Guilford & Zimmerman, 1949). The factors of
PRF impulsivity, MPQ cautious and methodical, and GZTS carefree, serious
minded, and spontaneous are correlated and themselves are well fit by a two-
factor solution of cautious versus spontaneous and methodical versus
disorganized (Parker et al., 1993). In a joint analysis of the EPQ, the I7, and
a German version of Cloninger's Tridimensional Personality Questionnaire
(Cloninger, 1987), Impulsivity from the I7 loads heavily on a factor defined by
EPQ-E, TPQ-Novelty Seeking, and TPQ-Reward Dependence (Weyers,
Krebs, & Janke, 1995). As would be expected by Gray's model, EPI-Imp has
higher corrations with Negative Affect than Positive Affect, a pattern that is
the reverse of the correlations of EPI-Soc with affect (Emmons & Diener,
1986). This pattern of relationships interacts with neuroticism such that for
high neurotics impulsivity and sociability are strongly related to positive affect,
while for low neurotics impulsivity is equally related to positive and negative
affect (McFatter, 1994).
Although analyses of the many separate scales of impulsivity indicate that it
is a multidimensional construct, a clear demonstration of this comes from an
examination of the multivariate structure of a pooled set of 378 items taken
198 Personality

from the existing impulsivity scales and measures of Barratt, Cattell, Eysenck,
Guilford, Jackson, Kagan, and Zuckerman (Gerbing, et al. 1987). From this
large set of items, 15 oblique first-order factors and three broad and correlated
second-order factors were identified. From the 12 first-order factors of self-
report (impulsive, energetic, quick decision making, thrill seeking, avoiding
planning, impulsive purchases, unreflective, avoids complexity, distractible,
restless, impatient, and happy go lucky), three second-level factors of
spontaneous, not persistent, and carefree were formed.

3. IMPULSIVITY IN CHILDREN AND ADULTS

Whether because of genetic effects (Eaves & Eysenck, 1975; Tellegen et al.,
1988) or the complex interplay of early temperament shaping the subsequent
environment (Caspi, 1993; Derryberry & Rothbart, 1988), impulsive children
tend to be more impulsive adults (Caspi & Silva, 1995; af Klinteberg,
Magnusson, & Schalling, 1989). Childhood impulsivity is of central concern to
theories of delinquency (John et al., 1994) and Attention Deficit-Hyperactivity
Disorder (ADHD) (Barkley, 1997; Douglas, 1972), but until recently there has
been little contact between theories of child and adult impulsivity. A few
theoretical discussions of ADHD consider the inability to sustain attention as a
sign of low arousal that results in a need for stimulation (S. S. Zentall & T. R.
Zentall, 1983), with some awareness of the adult literature on arousal seeking
(Eysenck, 1967) and stimulation seeking (Zuckerman, 1994) and more recent
theories of temperament have started to integrate the structure of childhood
temperament with adult personality structure (Eysenck, 1991b, 1994;
Halverson, Kohnstamm, & Martin, 1994; Rothbart, 1991; Strelau, 1991;
Strelau & Angleitner, 1991). A commonly used measure of reflection versus
impulsivity with children and adults is the Matching Familiar Figures Test
(Kagan 1966; Messer, 1976) that has low to zero correlations with paper and
pencil measures of impulsivity (Gerbing et al., 1987; Helmers et al., 1995).
Ratings of childhood temperament using the EASI (Buss and Plomin, 1975) or
Q-sort methodology show more favorable promise (John et al., 1994).

4. IMPULSIVITY AND AROUSAL

4.1 Introduction
For many of us interested in the relationship of personality with individual
differences in cognitive performance, the theoretical framework proposed in
the Biological Basis of Personality (Eysenck, 1967) acted as a navigational map
for our explorations (Revelle, 1995). Although framed in terms of extraversion
rather than impulsivity, the arousal model provided a common starting point.
 Extraversion and impulsivity: The lost dimension? 199

That much of the subsequent research involved impulsivity rather than

extraversion is a tribute to the theoretical richness of the original model and
the programmatic research that Eysenck inspired. The basic assumptions were:
(1) introverts are more aroused than extraverts; (2) stimulation increases arousal;
(3) arousal is curvilinearly related to performance; (4) the optimal level of arousal
for a task is negatively related to task difficulty; and (5) arousal is curvilinearly
related to hedonic tone. Assumption 1 was based upon many studies associating
EPI-E with (low) physiological arousal (Eysenck, 1967); assumptions 3 and 4
were based upon the Yerkes-Dodson Law (Yerkes & Dodson, 1908) and
subsequent support for it by Broadhurst (1959). Assumption 5 was founded on
Berlyne's discussion of curiosity and arousal (1960). Based upon assumptions 1-
4, it can be predicted that introverts should perform better than extraverts under
low levels of stimulation but should perform less well at high levels of stimulation.
Similarly, assumptions 1, 2, and 5 lead to the prediction that extraverts should
seek out more stimulation than introverts.
As the Eysencks tended to emphasize impulsivity in the PEN model as part
of the P scale, and to redefine the measurement of E within the EPQ, a
number of reanalyses of prior relationships of extraversion with behavioral,
physiological, and cognitive measures started to appear. Frequently, what had
previously been reported as relationships between extraversion and arousal
were found to hold for the EPI-E scale but not for the EPQ-E scale and in fact
to hold for the EPI-Imp but not the EPI-Soc subscales of the EPI.
EEG alpha activity shows a complex relationship with extraversion. Under
moderately stimulating conditions, extraverts are less aroused than introverts
(Gale, 1981), although this effect seems to be due to impulsivity (O'Gorman &
Lloyd, 1987; Stenberg, 1992, 1994), it is still a weak relationship (Matthews &
Amelang, 1993).
Impulsivity is related to the augmentation of the evoked potential response,
at least when recordings are taken at the vertex and frontal locations (Carrillo-
de-la-Pena & Barratt, 1993). This result is consistent with prior findings
relating ERP augmentation to sensation seeking and disinhibited behavior
(Barratt, Pritchard, Faulk, & Brandt, 1987) and demonstrates the need for
careful parametric specification of recording sites.
Impulsivity and caffeine-induced arousal have cross-over interactive effects
on skin conductance measures of arousal such that with placebo low impulsives
have higher SCL than high impulsives, but although both groups increase with
caffeine, the high impulsives now have higher levels of SCL (Smith, Rypma, &
Wilson, 1981). This effect was not found for measures of sociability.

4.2 Impulsivity and stimulation preference

A conclusion from the Biological Basis of Personality (Eysenck, 1967) is that
introverts are chronically more aroused than extraverts. This, the assumption
200 Personality

that there is an optimal level of stimulation leads to the prediction that

extraverted behavior represents a greater stimulus hunger on the part of the
less aroused extraverts. In a reanalysis of a previous result which had shown
that extraverts prefer to study in noisier conditions than introverts, Campbell
(1983) found that this effect was due to impulsivity rather than sociability. In a
further analysis of the relationship of personality to tolerance for noise,
Campbell (1992) found that when controlling for neuroticism there was a
stronger effect for impulsivity than for sociability. Campbell and Heller (1987)
found that both sociability and impulsivity related to Zuckerman's sensation
seeking scale and that sociability had much higher correlations with the
Meyer-Briggs Temperament Inventory Introversion-Extraversion measure
than impulsivity. Presumably reflecting a need for stimulation, more impulsive
athletes prefer "explosive" sports while less impulsive athletes prefer
"endurance" sports (Svebak & Kerr, 1989).
An important behavioral finding is the greater number of traffic violations
and accidents for high impulsives than for low impulsives (Loo, 1979). In an
examination of preference for bright (red, yellow) versus dull (blue, green)
colors, Zuber and Ekehammar (1988) found that impulsivity interacted with
time of day such that high impulsives preferred the bright colors in the
morning but not in the evening, while low impulsives preferred brighter colors
later in the day. Such an interaction of impulsivity with time of day in
preferences is consistent with the finding that EPI-Imp is more correlated than
EPI-Soc with preferred time of day for rising and retiring (morningness-
eveningness; Neubauer, 1992).
The association of impulsivity with sensation seeking has led to alternative
factor analytic rotations of three- and five-dimensional solutions for
personality taxonomies with "impulsive-unsocialized sensation seeking"
proposed as one of the fundamental dimensions of personality (Zuckerman,
1994; Zuckerman et al., 1993; see chapter 1 in this volume). In several domains
of risky behavior, high impulsives give lower estimates of personal risk and
have higher rates of engaging in risky behavior than low impulsives (Horvath &
Zuckerman, 1993). Impulsivity as measured by the I7 in combination with
markers for unsocialized sensation seeking discriminates between a group of
prisoners and control and prosocial or risky sport enthusiasts (Goma-i-
Freixanet, 1995).

4.3 Impulsivity and conditioning

An early demonstration of the importance of impulsivity and task parameters
in conditioning was the finding that low impulsives showed more rapid
eyeblink conditioning than high impulsives under conditions of stimulation but
that this effect reversed under higher levels of stimulation. (Eysenck & Levey,
1972). This effect did not hold for sociability. After several failures to replicate
 Extroversion and impulsivity: The lost dimension? 201

this result using broader measures of impulsivity, Frcka and Martin (1987)
reported that impulsivity in the narrow sense (essentially the impulsivity items
from the EPI) did interact with stimulus intensity but broader measures from
the I5 and I7 did not. Their article is an excellent review of the problems
encountered when presumed psychometric refinements lead to experimental
difficulties (see chapter 16 in this volume).
Several tests of Gray's hypothesis (1982, 1987, 1990) that impulsivity is
related to sensitivity to cues for reward and that anxiety is related to sensitivity
to cues for punishment have used conditioning paradigms. In a go-no go
discrimination task where type of response was crossed with rewards or
punishments, impulsivity and anxiety interact to predict response frequency
(Zinbarg & Revelle, 1989). Similar results showing that anxiety and impulsivity
provide a better fit to conditioning data than neuroticism and extraversion
have been reported by Corr, Pickering, and Gray (1995) and by Diaz and
Pickering (1993) with negative results by Pickering, Diaz, and Gray (1995).
Newman and his colleagues have tested Gray's hypothesis by using the
combination of neuroticism and extraversion as a surrogate for impulsivity
(Bachorwski & Newman 1985, 1990; Newman, 1987; Wallace & Newman,
1980). Note that although framed in terms of Gray's model of impulsivity and
anxiety, tests of his theory using E+N are not direct tests of impulsivity.

4.4 Impulsivity and cognitive performance

In addition to predicting correlations of extraversion with biological markers of
arousal and making the predictions that people who say they enjoy lively
parties and doing things quickly without stopping to think actually do so, the
Biological Basis of Personality made predictions that allowed for an integration
of personality and experimental psychology. Applying assumptions 1-4 led to
predictions of how extraversion would combine with situationally induced
arousal to affect performance. Specifically, introverts were expected to perform
better than extraverts in situations that induced low arousal but to perform less
well in situations that induced high arousal.
Seeming support for this prediction was the finding that time pressure and
caffeine-induced arousal hinders the performance of introverts but facilitates
that of extraverts on a test similar to the verbal portion of the Graduate Record
Examination (GRE) (Revelle, Amaral, & Turriff, 1976). Introverts performed
best under conditions of no time pressure and no caffeine and their per-
formance deteriorated with the introduction of time pressure and deteriorated
even more with the combination of time pressure plus 200 mg of caffeine.
Extraverts, on the other hand, performed worst in the low stress condition and
best in the time pressure plus caffeine condition. However, in a conceptual
replication of this study using three rather than two levels of caffeine, these
202 Personality

effects were only consistent for EPI-E and not for EPQ-E (Gilliland, 1976).
The difference turned out to be that the effects were due to the impulsivity
items on the EPI-E scale.
Subsequent investigation showed that while caffeine reliably increased GRE
performance for high impulsives and hindered it for low impulsives, these
effects were only true in the morning and in fact reversed in the evening
(Revelle et al., 1980). In a set of five new experiments and reanalyses of the
Revelle et al. (1976) and Gilliland (1976) experiments there was a consistent,
although complex triple interaction of impulsivity, caffeine induced arousal,
and time of day. The pattern with sociability was much less consistent.
In an independent replication and extension of these findings, Matthews
(1987) measured extraversion, impulsivity, and sociability using scales from
Cattell's 16PF and used self-reported arousal rather than manipulated arousal.
A triple interaction of impulsivity, self reported arousal, and time of day was
remarkably similar to the Revelle et al. (1980) results: low aroused low
impulsives did better than high aroused low impulsives or low aroused high
impulsives in the morning but this result reversed in the evening. Subsequent
investigations by Matthews and his colleagues have shown that the time of day
by arousal by extraversion interaction is the prototypical result although the
relative contributions of sociability and impulsivity seem to be inconsistent
(Matthews, Davies, & Lees, 1990; Matthews, Jones, & Chamberlin, 1989).
Interactive effects of impulsivity, caffeine, and time of day have also been
reported by Smith et al. (1991).
Besides demonstrating that the arousal effects previously attributable to
extraversion were more likely to be associated with impulsivity, these results
called into question the basic assumption that extraverts were in fact always
less aroused than introverts. The time of day results suggested that stable
arousal differences could not account for the greater stimulation seeking of
impulsives and extraverts, for otherwise, why would not extraverts be introverts
at night? (Note, however, that Larsen (1985) reports diurnal variation in
measures of arousal to be more related to sociability than to impulsivity.)
Revising Eysenck's first assumption to be that (la) low impulsives are more
aroused than high impulsives and (Ib) this relationship reversed in the evening
led to a series of studies showing consistent patterns of impulsivity by caffeine-
induced arousal interactions (reviewed in Revelle, 1989). Caffeine interacts
with memory load requirements on a proof-reading task such that it facilitates
performances for high impulsives but not for low impulsives when memory
load is high (Anderson & Revelle, 1982). Caffeine facilitated performance on a
complex visual scanning task for high impulsives but did not for low impulsives
(Anderson & Revelle, 1983). In the morning, in a superspan memory task that
required sustained attention, high impulsives showed a bigger decline in
performance across trials than low impulsives and the decline in performance
 Extraversion and impulsivity: The lost dimension? 203

was minimized by caffeine (Bowyer, Humphreys, & Revelle, 1983). Although

this result can be replicated in the morning, it reverses in the evening
(Anderson & Revelle, 1994).
Concerned that the evidence for the Yerkes-Dodson Law was based upon
aggregation of between subjects effects and might not hold within subjects,
Anderson (1994) examined the effects of five levels of caffeine on simple and
complex performance tasks for high and low impulsives. Consistent with prior
studies, the between subjects data showed a quadratic effect (inverted U
pattern) for the low impulsives and a linear effect (increasing pattern) for the
high impulsives. Applying an elegant analysis to the within subject patterns,
Anderson concluded that the inverted U pattern occurred at the individual
level with a reliable frequency and was not an artifact of data aggregation.
Impulsivity is an important component of the model of how personality traits
interact with situational states to affect cognitive performance outlined by
Humphreys and Revelle (1984) and subsequent developments of that model
(Anderson, Revelle, & Lynch, 1989; Revelle, 1989,1993; Revelle & Anderson,
1992; Revelle, Anderson, & Humphreys, 1987). Impulsivity systematically
interacts with time of day in its effects upon cognitive performance. The most
parsimonious interpretation of these results is that low impulsives are more
aroused than high impulsives early in the day but are less aroused than high
impulsives in the evening.
The relationship of impulsivity to performance needs to be considered at
multiple levels of analysis and not just in terms of the arousal-mediated effects
(Revelle, 1987). Perhaps because they do well in time stressed situations, or
perhaps because they are more sensitive to cues for rewards than to
punishments, impulsives are more likely to engage in behaviors that put
them in highly arousing situations and to adopt lifestyles that are focused on
rewards and not concerned with the possible negative consequences. Such
stylistic choices are not directly arousal related but do lead to lifelong
differences in preferences that are modified only slightly by moment to
moment or day to day shifts in arousal. How differences in impulsivity affect
performance is also a function of other personality traits and abilities.
Intelligence and anxiety should act as control mechanism to moderate the
quick tempo and reward sensitivity of the high impulsive. Less able and less
anxious impulsives should be much more likely to exhibit the problematic
behavioral disorders associated with impulsivity than more intelligent and
anxious impulsives.

5. CONCLUSIONS

Impulsivity has long been seen as an important component of individual

differences. Its place in a multidimensional personality theory is, however, less
204 Personality

clear. Although its location in personality space has moved from a central part
of extraversion to a blend of neuroticism and psychoticism to a neglected part
of the lexical description of personality, impulsivity seems to show strong
biological and behavioral correlates. Is this because impulsivity is a blend of
constructs, each of which separately has a biological basis, or is impulsivity a
surface marker for an underlying biological system? Only time will help us to
resolve this question, and only then if the high quality of psychometric and
biologically driven research inspired by Hans Eysenck continues to examine
this important domain.

REFERENCES

Amelang, M., & Ullwer, U. (1991). Correlations between psychometric measures and
psychophysiological as well as experimental variables in studies of extraversion and
neuroticism. In J. Strelau & A. Angleitner (Eds.), Explorations in temperament:
International perspectives on theory and measurement (pp. 287-315). London: Plenum
Press.
Anderson, K. J. (1994). Impulsivity, caffeine, and task difficulty: A within-subjects test
of the Yerkes-Dodson law. Personality & Individual Differences, 16, 813-829.
Anderson, K. J., & Revelle, W. (1982). Impulsivity, caffeine, and proofreading: A test of
the Easterbrook hypothesis. Journal of Experimental Psychology: Human Perception &
Performance, 8, 614-624.
Anderson, K. J., & Revelle, W. (1983). The interactive effects of caffeine, impulsivity
and task demands on a visual search task. Personality & Individual Differences, 4, 127-
134.
Anderson, K. J., & Revelle, W. (1994). Impulsivity and time of day: Is rate of change in
arousal a function of impulsivity? Journal of Personality & Social Psychology, 67, 334-
344.
Anderson, K. J., Revelle, W., & Lynch, M. J. (1989). Caffeine, impulsivity, and memory
scanning: A comparison of two explanations for the Yerkes-Dodson Effect.
Motivation & Emotion, 13, 1-20.
Bachorowski, J. A., & Newman, J. P. (1985). Impulsivity in adults: Motor inhibition and
time-interval estimation. Personality & Individual Differences, 6, 133-136.
Bachorowski, J. A., & Newman, J. P. (1990). Impulsive motor behavior: Effects of
personality and goal salience. Journal of Personality & Social Psychology, 58, 512-518.
Barkley, R. A. (1997). Behavioral inhibition, sustained attention, and executive
functions: Constructing a unifying theory of ADHD. Psychological Bulletin, 121,
65-94.
Barratt, E. S. (1987). Impulsiveness and anxiety: information processing and
electroencephalograph topography. Journal of Research in Personality, 21, 453-463.
Barratt, E. S., & Patton, J. H. (1983). Impulsivity: Cognitive, behavioral, and
psychophysiological correlates. In M. Zuckerman (Eds.), Biological bases of sensation
seeking, impulsivity, and anxiety (pp. 77-116). Hillsdale, NJ: Lawrence Erlbaum
Associates.
 Extraversion and impulsivity: The lost dimension? 205

Barratt, E. S., Pritchard, W. S., Faulk, D. M., & Brandt, M. E. (1987). The relationship
between impulsiveness subtraits, trait anxiety, and visual N100 augmenting/reducing:
a topographic analysis. Personality and Individual Differences, 8, 43-51.
Berlyne, D. E. (1960). Conflict, arousal, and curiosity. New York: McGraw-Hill.
Bowyer, P. A., Humphreys, M. S., & Revelle, W. (1983). Arousal and recognition
memory: The effects of impulsivity, caffeine and time on task. Personality & Individual
Differences, 4, 41-49.
Broadhurst, P. L. (1959). The interaction of task difficulty and motivation: the Yerkes-
Dodson Law revived. Acta Psychologica, 16, 321-328.
Brunas-Wagstaff, J., Bergquist, A., & Wagstaff, G. F. (1994). Cognitive correlates of
functional and dysfunctional impulsivity. Personality & Individual Differences, 17, 289-
292.
Brunas-Wagstaff, J., Bergquist, A., Richardson, P., & Connor, A. (1995). The
relationships between functional and dysfunctional impulsivity and the Eysenck
personality questionnaire. Personality & Individual Differences, 18, 681-683.
Buss, A. H., & Plomin, R. (1975). A temperament theory of personality development. New
York: Wiley.
Campbell, J. B. (1983). Differential relationships of extraversion, impulsivity, and
sociability to study habits. Journal of Research in Personality, 17, 308-314.
Campbell, J. B. (1992). Extraversion and noise sensitivity: A replication of Dornic and
Ekehammar's study. Personality & Individual Differences, 13, 953-955.
Campbell, J. B., & Heller, J. F. (1987). Correlations of extraversion, impulsivity and
sociability with sensation seeking and MBTI-introversion. Personality & Individual
Differences, 8, 133-136.
Campbell, J. B., & Reynolds, J. H. (1984). A comparison of the Guilford and Eysenck
factors of personality. Journal of Research in Personality, 18, 305-320.
Carrigan, P. M. (1960). Extraversion-introversion as a dimension of personality: a
reappraisal. Psychological Bulletin, 57, 329-360.
Carrillo-de-la-Pena, M., T., & Barratt, E. S. (1993). Impulsivity and ERP augmenting/
reducing. Personality & Individual Differences, 15, 25-32.
Caspi, A. (1993). Why maladaptive behaviors persist: sources of continuity and change
across the life course. In D. Funder, R. D. Parke, C. Tomlinson-Keasey, & K.
Widman (Eds.), Studying lives through time (pp. 343-376) Washington, DC: American
Psychological Association.
Caspi, A., & Silva, P. A. (1995). Temperamental qualities at age three predict
personality traits in young adulthood: Longitudinal evidence from a birth cohort.
Child Development, 66, 486-498.
Cloninger, C. R. (1987). A systematic method for clinical description and classification
of personality variants—a proposal. Archives of General Psychiatry, 44, 573-588.
Corr, P. J., Pickering, A. D., & Gray, J. A. (1995). Personality and reinforcement hi
associative and instrumental learning. Personality & Individual Differences, 19, 47-71.
Costa, P. T., & McCrae, R. R. (1992). Four ways five factors are basic. Personality &
Individual Differences, 13, 653-665.
Derryberry, D., & Rothbart, M. K. (1988). Arousal, affect, and attention as components
of temperament. Journal of Personality and Social Psychology, 55, 958-966.
Diaz, A., & Pickering, A. D. (1993). The relationship between Gray's and Eysenck's
personality spaces. Personality & Individual Differences, 15, 297-305.
206 Personality

Dickman, S. J. (1990). Functional and dysfunctional impulsivity: Personality and

cognitive correlates. Journal of Personality & Social Psychology, 58, 95-102.
Dickman, S. J., & Meyer, D. E. (1988). Impulsivity and speed-accuracy tradeoffs in
information processing. Journal of Personality & Social Psychology, 54, 274-290.
Digman, J. M. (1994). Child personality and temperament: does the five factor model
embrace both domains. In C. F. Halverson Jr., A. Kohnstamm, & R. P. Martin
(Eds.), The developing structure of temperament and personality from infancy to
adulthood (pp. 323-338). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Douglas, V. I. (1972). Stop, look and listen: The problem of sustained attention and
impulse control in hyperactive and normal children. Canadian Journal of Behavioral
Science, 4, 259-282.
Eaves, L., & Eysenck, H. J. (1975). The nature of extraversion: A genetical analysis.
Journal of Personality and Social Psychology, 32, 102-112.
Emmons, R. A., & Diener, E. (1986). Influence of impulsivity and sociability on
subjective well-being. Journal of Personality & Social Psychology, 50, 1211-1215.
Eysenck, H. J. (1947). Dimensions of personality. London: Routledge & Kegan Paul.
Eysenck, H. J. (1956). The questionnaire measurement of neuroticism and extraversion.
Revista de Psicologia, 50, 113-140.
Eysenck, H. J. (1958). A short questionnaire for the measurement of two dimensions of
personality. Journal of Applied Psychology, 42, 14-17.
Eysenck, H. J. (1959). Manual for the Maudsley Personality Inventory. London:
University of London Press.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Thomas.
Eysenck, H. J. (1977). Personality and factor analysis: A reply to Guilford. Psychological
Bulletin, 84, 405-411.
Eysenck, H. J. (1987). The place of anxiety and impulsivity in a dimensional framework.
Journal of Research in Personality, 21, 489-492.
Eysenck, H. J. (1990). Biological dimensions of personality. In L. A. Pesvin (Ed.),
Handbook of personality: theory and research (pp. 244-276). New York: Guilford.
Eysenck, H. J. (1991a). Dimensions of personality: 16, 5 or 3? Criteria for a taxonomic
paradigm. Personality & Individual Differences, 12, 773-790.
Eysenck, H. J. (1991b). Dimensions of personality: the biosocial approach to
personality. In J. Strelau & A. Angleitner (Eds.), Explorations in temperament:
international perspectives in theory and measurement London: Plenum.
Eysenck, H. J. (1992). A reply to Costa and McCrae: P or A and C—the role of theory.
Personality & Individual Differences, 13, 867-868.
Eysenck, H. J. (1994). The Big Five or Giant Three: Criteria for a paradigm. In C. F.
Halverson Jr., A. Kohnstamm, & R. P. Martin (Eds.), The developing structure of
temperament and personality from infancy to adulthood. Hillsdale, NJ: Lawrence
Erlbaum Associates, Inc.
Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual differences: a natural
science approach. New York: Plenum Press.
Eysenck, H. J., & Eysenck, S. B. G. (1964). Eysenck Personality Inventory. San Diego,
CA: Educational and Industrial Testing Service.
Eysenck, H. J., & Eysenck, S. B. G. (1967). On the unitary nature of extraversion. Acta
Psychologica, 26, 383-390.
 Extraversion and impulsivity: The lost dimension? 207

Eysenck, H. J., & Eysenck, S. B. G. (1969). Personality structure and measurement. San
Diego, CA: Robert R. Knapp.
Eysenck, H. J., & Eysenck, S. B. G. (1975) Manual of the Eysenck Personality
Questionnaire. London: Hodder & Stoughton
Eysenck, S. B. G., & Eysenck, H. J. (1963). On the dual nature of extraversion. British
Journal of Social and Clinical Psychology, 2, 46-55.
Eysenck, S. B. G., & Eysenck, H. J. (1969). The dual nature of extraversion. In H. J.
Eysenck & S. B. G. Eysenck (Eds.), Personality structure and measurement (pp. 141-
149). San Diego, CA: Robert Knapp.
Eysenck, S. B. G., & Eysenck, H. J. (1977). The place of impulsiveness in a dimensional
system of personality description. British Journal of Social & Clinical Psychology, 16,
57-68.
Eysenck, S. B. G., & Eysenck, H. J. (1978). Impulsiveness and venturesomeness: Their
position in a dimensional system of personality description. Psychological Reports, 43,
1247-1255.
Eysenck, S. B. G., Eysenck, H. J., & Barrett, P. (1985). A revised version of the
Psychoticism scale. Personality & Individual Differences, 6, 21-29.
Eysenck, H. J., & Levey, A. (1972). Conditioning, introversion-extraversion and the
strength of the nervous system. In V. D. Nebylitsyn & J. A. Gray (Eds.), Biological
bases of individual behavior. New York: Academic Press.
Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for
impulsiveness, venturesomeness and empathy in adults. Personality and Individual
Differences, 6, 613-619.
Fowles, D. C. (1987). Application of a behavioral-theory of motivation to the concepts
of anxiety and impulsivity. Journal of Research in Personality, 21, 417-435.
Frcka, G., & Martin, I. (1987). Is there—or is there not—an influence of impulsiveness
on classic eyelid conditioning. Personality and Individual Differences, 8, 241-252.
Gale, A. (1981). EEG studies of introversion-extraversion: what's the next step? In R.
Lynn (Ed.), Dimensions of Personality: Papers in honour of H.J. Eysenck. Oxford:
Pergamon.
Gerbing, D. W., Ahadi, S. A., & Patton, J. H. (1987). Toward a conceptualization of
impulsivity: Components across the behavioral and self-report domains. Multivariate
Behavioral Research, 22, 357-379.
Gilliland, K. (1976) The interactive effect of introversion-extraversion with caffeine induced
arousal on verbal performance. Unpublished doctoral dissertation, Northwestern
University.
Goldberg, L. R. (1990). An alternative "description of personality": the big-five factor
structure. Journal of Personality and Social Psychology, 59, 1216-1229.
Goma-i-Freixanet, M. (1995). Prosocial and antisocial aspects of personality. Personality
& Individual Differences, 19, 125-134.
Gray, J. A. (1972). The psychophysiological basis of introversion-extraversion: a
modification of Eysenck's theory. In V. D. Nebylitsyn & J. A. Gray (Eds.), The
biological basis of individual behavior. New York: Academic Press.
Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.),
A model for personality. Berlin: Springer-Verlag.
Gray, J. A. (1982). Neuropsychological Theory of Anxiety: An investigation of the septal-
hippocampal system. Cambridge: Cambridge University Press.
208 Personality

Gray, J. A. (1987). Perspectives on anxiety and impulsivity: a commentary. Journal of

Research in Personality, 21, 493-509.
Gray, J. A. (1990). Brain systems that mediate both emotion and cognition. Cognition
and Emotion, 4, 269-288.
Gray, J. A. (1991). The neuropsychology of temperament. In J. Strelau & A. Angleitner
(Eds.), Explorations in temperament: international perspectives on theory and
measurement (pp. 105-128). London: Plenum.
Gray, J. A. (1994). Framework for a taxonomy of psychiatric disorder. In S. H. M. v.
Goozen, N. E. v. Poll, & J. Sergeant (Eds.), Emotions: Essays on emotion theory (pp.
29-59). Hillsdale, NJ: Erlbaum.
Guilford, J. P. (1975). Factors and factors of personality. Psychological Bulletin, 82, 802-
814.
Guilford, J. P. (1977). Will the real factor of extraversion please stand up? A reply to
Eysenck. Psychological Bulletin, 84, 412-416.
Guilford, J. P., & Zimmerman, W. S. (1949). The GuUford-Zimmerman Temperament
Survey: Manual of instructions and interpretations. Beverly Hills, CA: Sheridan Supply
Co.
Hah/erson, C. F. Jr., Kohnstamm, G. A., & Martin, R. P. (Eds.). (1994). The developing
structure of temperament and personality from infancy to adulthood. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Helmers, K. F., Young, S. N., & Pihl, R. O. (1995). Assessment of measures of
impulsivity in healthy male volunteers. Personality & Individual Differences, 19, 927-
935.
Heymans, G. (1929). Inleidung in speciale psychologic. Haarlem: Bohm.
Hofstee, W. K. B., de Raad, B., & Goldberg, L. R. (1992). Integration of the big five and
circumplex approaches to trait structure. Journal of Personality and Social Psychology,
63, 146-163.
Horvath, P., & Zuckerman, M. (1993). Sensation seeking, risk appraisal, and risky
behavior. Personality & Individual Differences, 14, 41-52.
Humphreys, M. S., & Revelle, W. (1984). Personality, motivation, and performance: A
theory of the relationship between individual differences and information processing.
Psychological Review, 91, 153-184.
Jackson, D. N. (1984). Personality Research Form manual. Port Huron, MI: Research
Psychologists Press.
John, O. P. (1990). The "Big Five" factor taxonomy: Dimensions of personality in the
natural language and in questionnaires. In L. A. Pervin (Ed.), Handbook of
personality theory and research. New York: Guilford Press.
John, O. P., Caspi, A., Robins, R. W., Moffitt, T. E., et al. (1994). The "little five":
Exploring the nomological network of the five-factor model of personality in
adolescent boys. Child Development, 65, 160-178.
Kagan, J. (1966). Reflection-impulsivity: The generality and dynamics of conceptual
tempo. Journal of Abnormal Psychology, 71, 17-24.
Kassenbaum, C. G., Couch, A. S., & Slater, P. (1959). The factorial dimensions of the
MMPI. Journal of Consulting Psychology, 23, 226-236.
af Klinteberg, B., Magnusson, D., & Schalling, D. (1989). Hyperactive behavior in
childhood and adult impulsivity: a longitudinal study of male subjects. Personality and
Individual Differences, 10, 43-50.
 Extroversion and impulsivity: The lost dimension? 209

Larsen, R. J. (1985). Individual differences in circadian activity rhythm and personality.

Personality & Individual Differences, 6, 305-311.
Loo, R. (1979). Role of primary personality factors in the perception of traffic signs and
driver violations and accidents. Accident Analysis & Prevention, 11, 125-127.
Luengo, M. A., Carrillo-de-la-Pena, M. T., & Otero, J. M. (1991). The components of
impulsiveness: A comparison of the 17 impulsiveness questionnaire and the Barratt
impulsiveness scale. Personality and Individual Differences, 12, 657-667.
Matthews, G. (1987). Personality and multidimensional arousal: A study of two
dimensions of extraversion. Personality & Individual Differences, 8, 9-16.
Matthews, G., & Amelang, M. (1993). Extraversion, arousal theory and performance: A
study of individual differences in the EEG. Personality & Individual Differences, 14,
347-363.
Matthews, G., Davies, D. R., & Lees, J. L. (1990). Arousal, extroversion, and individual-
differences in resource availability. Journal of Personality and Social Psychology, 59,
150-168.
Matthews, G., Jones, D. M., & Chamberlain, A. G. (1989). Interactive effects of
extraversion and arousal on attentional task-performance—multiple resources or
encoding processes. Journal of Personality and Social Psychology, 56, 629-639.
McCrae, R. R., & Costa, P. T. (1985). Comparison of EPI and psychoticism scales with
measures of the five-factor model of personality. Personality & Individual Differences,
6, 587-597.
McFatter, R. M. (1994). Interactions in predicting mood from extraversion and
neuroticism. Journal of Personality & Social Psychology, 66, 570-578.
Messer, S. B. (1976). Reflection-impulsivity: A review. Psychological Bulletin, 83, 1026-
1052.
Murray, H. A. (1938). Explorations in personality: A clinical and experimental study of fifty
men of college age. New York: Oxford University Press.
Neubauer, A. C. (1992). Psychometric comparison of two circadian rhythm
questionnaires and their relationship with personality. Personality & Individual
Differences, 13, 125-131.
Newman, J. P. (1987). Reaction to punishment in extraverts and psychopaths:
implications for the impulsive behavior of disinhibited individuals. Journal of
Research in Personality, 21, 464-480.
O'Gorman, J. G., & Lloyd, J. E. M. (1987). Extraversion, impulsiveness, and EEG alpha
activity. Personality and Individual Differences, 8, 169-174.
Parker, J. D. A., Bagby, R. M., & Webster, C. D. (1993). Domains of the impulsivity
construct: A factor analytic investigation. Personality & Individual Differences, 15,
267-274.
Pickering, A. D., Diaz, A., & Gray, J. A. (1995). Personality and reinforcement: An
exploration using a maze-learning task. Personality & Individual Differences, 18, 541-
558.
Rawlings, D. (1984). The correlation of EPQ Psychoticism with two behavioural
measures of impulsivity. Personality & Individual Differences, 5, 591-594.
Revelle, W. (1986). Motivation and efficiency of cognitive performance. In D. R. Brown
& J. Veroff (Eds.) Frontiers of motivational psychology: Essays in honor of J. W.
Atkinson (pp. 105-131). Berlin: Springer.
210 Personality

Revelle, W. (1987). Personality and motivation: Sources of inefficiency in cognitive

performance. Journal of Research in Personality, 21, 436-452.
Revelle, W. (1989). Personality, motivation, and cognitive performance. In P.
Ackerman, R. Kanfer, & R. Cudeck (Eds.), Learning and individual differences:
abilities, motivation, and methodology (pp. 297-341). Hillsdale, NJ: Lawrence
Erlbaum Associates.
Revelle, W. (1993). Individual differences in personality and motivation: 'Non-
cognitive' determinants of cognitive performance. In A. Baddeley & L. Weiskrantz
(Eds.), Attention: selection, awareness and control: a tribute to Donald Broadbent
(pp. 346-373). Oxford: Oxford University Press.
Revelle, W. (1995). Personality processes. Annual Review of Psychology, 46, 295-328.
Revelle, W., Amaral, P., & Turriff, S. (1976). Introversion/extroversion, time stress, and
caffeine: Effect on verbal performance. Science, 192, 149-150.
Revelle, W., & Anderson, K. J. (1992). Models for the testing of theory. In A. Gale &
M. W. Eysenck (Eds.), Handbook of individual differences: Biological perspectives
(pp. 81-113). Chichester, U.K.: John Wiley & Sons.
Revelle, W., Anderson, K. J., & Humphreys, M. S. (1987). Empirical tests and
theoretical extensions of arousal-based theories of personality. In J. Strelau & H. J.
Eysenck (Eds.), Personality dimensions and arousal (pp. 17-36). London: Plenum.
Revelle, W., Humphreys, M. S., Simon, L., & Gilliland, K. (1980). The interactive effect
of personality, time of day, and caffeine: A test of the arousal model. Journal of
Experimental Psychology: General, 109, 1-31.
Rocklin, T., & Revelle, W. (1981). The measurement of extraversion: A comparison of
the Eysenck Personality Inventory and the Eysenck Personality Questionnaire. British
Journal of Social Psychology, 20, 279-284.
Roger, D., & Morris, J. (1991). The internal structure of the EPQ scales. Personality &
Individual Differences, 12, 759-764.
Romero, E., Luengo, M. A., Carrillo-de-la-Pena, M. T., Teresa, M., Otero, L., & Jose,
M. (1994). The Act Frequency Approach to the study of impulsivity. European
Journal of Personality, 8, 119-133.
Rothbart, M. K. (1991). Temperament: a developmental framework. In J. Strelau & A.
Angleitner (Eds.), Explorations in temperament: international perspectives in theory
and measurement J(pp. 61-74). London: Plenum.
Schalling, D., & Asberg, M. (1985). Biological and psychological correlates of
impulsiveness and monotony avoidance. In J. Strelau, F. H. Farley, & A. Gale
(Eds.), The biological bases of personality and behavior, VoL 1: Theories, measurement
techniques, and development. Washington, DC: Hemisphere Publishing Corp/Harper
& Row Publishers.
Schalling, D., Edman, G., & Asberg, M. (1983). Impulsive cognitive style and inability to
tolerate boredom: Psychobiological studies of temperamental variability. In M.
Zuckerman (Ed.), Biological bases of sensation seeking, impulsivity, and anxiety, (pp.
123-145). Hillsdale, NJ: Lawrence Erlbaum Associates.
Smith, A. P., Rusted, J. M., Savory, M., Eaton, W. P., et al. (1991). The effects of
caffeine, impulsivity and time of day on performance, mood and cardiovascular
function. Journal of Psychopharmacology, 5, 120-128.
 Extraversion and impulsivity: The lost dimension? 211

Smith, B. D., Rypma, C. B., & Wilson, R. J. (1981). Dishabituation and spontaneous
recovery of the electrodermal orienting response: Effects of extraversion, impulsivity,
sociability, and caffeine. Journal of Research in Personality, 15, 233-240.
Sparrow, N. H., & Ross, J. (1964). The dual nature of extraversion: a replication.
Australian Journal of Psychology, 16, 214-218.
Stenberg, G. (1992). Personality and the EEG: Arousal and emotional arousability.
Personality & Individual Differences, 13(10), 1097-1113.
Stenberg, G. (1994). Extraversion and the P300 in a visual classification task. Personality
& Individual Differences, 16, 543-560.
Strelau, J. (1991). Renaissance in research on temperament: where to? In J. Strelau &
A. Angleitner (Eds.), Explorations in temperament: international perspectives on theory
and measurement (pp. 337-358). London: Plenum.
Strelau, J., & Angleitner, A. (Eds.). (1991). Explorations in temperament: international
perspectives on theory and measurement. London: Plenum.
Svebak, S., & Kerr, J. H. (1989). The role of impulsivity in preference for sports.
Personality & Individual Differences, 10, 51-58.
Tellegen, A. (1982). Brief manual for the Multidimensional Personality Questionnaire.
Minneapolis, MS: University of Minnesota.
Tellegen, A. (1985). Structures of mood and personality and their relevance to assessing
anxiety, with an emphasis upon self report. In A. H. Tuma & J. Maser (Eds.), Anxiety
and the anxiety disorders (pp. 681-706). Hillsdale, NJ: Lawrence Erlbaum Associates.
Tellegen, A., Lykken, D. T., Bouchard, T. J., Wilcox, K. J., Rich, S., & Segal, N. L.
(1988). Personality similarity in twins reared apart and together. Journal of Personality
and Social Psychology, 54, 1031-1039.
Van der Werff, J. J. (1985). Heyman's temperamental dimensions in personality
research. Journal of Research in Personality, 19, 279-287.
Van der Werff, J. J., & Verster, J. (1987). Heymans' temperamental dimensions
recomputed. Personality & Individual Differences, 8, 271-276.
Wallace, J. F., & Newman, J. P. (1990). Differential effects of reward and punishment
cues on response speed in anxious and impulsive individuals. Personality & Individual
Differences, 11, 999-1009.
Weyers, P., Krebs, H., & Janke, W. (1995). Reliability and construct validity of the
German version of Cloninger's Tridimensional Personality Questionnaire. Personality
& Individual Differences, 19, 853-861.
Wolfe, R. N., & Kasmer, J. A. (1988). Type versus trait: Extraversion, impulsivity,
sociability, and preferences for cooperative and competitive activities. Journal of
Personality & Social Psychology, 54, 864-871.
Yerkes, R. M., & Dodson, J. D. (1908). The relation of strength of stimuli to rapidity of
habit-information. Journal of Comparative Neurology and Psychology, 18, 459-482.
Zentall, S. S., & Zentall, T. R. (1983). Optimal stimulation: A model of disordered
activity and performance in normal and deviant children. Psychological Bulletin, 94,
446-471.
Zinbarg, R., & Revelle, W. (1989). Personality and conditioning: A test of four models.
Journal of Personality & Social Psychology, 57, 301-314.
Zuber, I., & Ekehammar, B. (1988). Personality, time of day and visual perception:
Preferences and selective attention. Personality & Individual Differences, 9, 345-352.
212 Personality

Zuckerman, M. (1991). Psychobiology of personality. Cambridge: Cambridge University

Press.
Zuckerman, M. (1994). Behavioral expressions and biosocial bases of sensation seeking.
Cambridge University Press; New York.
Zuckerman, M., Kuhlman, D. M., Joireman, J., Teta, P., & Kraft, M. (1993). A
comparison of three structural models for personality: The Big Three, the Big Five,
and the Alternative Five. Journal of Personality & Social Psychology, 65, 757-768.
PART II

Intelligence
Introduction: Hans Eysenck and the study of
intelligence

A. R. Jensen

In understanding Eysenck's contributions to this field, it may help to sketch

some of the historical background into which he was first introduced as a
student of psychology upon his arrival in London in 1935, and which, I believe,
must have greatly influenced his subsequent career, especially as it involves the
study of intelligence.
In a recent interview for a popular magazine, Eysenck stated that one of the
first topics in psychology that interested him was intelligence. The reason he
gave for this attraction was that intelligence was one of the first human traits
for which psychologists could actually measure individual differences with
some accuracy.
The idea that measurement, rather than subjective impressions, is a central
requirement for the development of psychology as a branch of empirical
natural science has been an abiding conviction of Eysenck's from the very
beginning of his long and distinguished career. This value probably followed
naturally from the fact that, as an emigre from Hitler's Germany, he had gone
to the University of London hoping to major in physics. For certain technical
reasons, however, he was not permitted to enroll as a major in that field. So, at
that time, psychology was the only scientific subject open to him, and he had
little option but to have that as his major, to scientific psychology's good
fortune.
It remains uncertain whether Eysenck's deeply ingrained proclivity for
measurement, quantitative analysis, and a wholly natural science approach to
psychology, was innately intrinsic to his intellect and personality or was mainly
imposed by his education in the German gymnasium, or later at University
College, London. Most likely it was just a case of a genuine compatibility
between Eysenck's scientific nature and the fortunate happenstance of the
particular nurture afforded him by his initiation into a type of psychology that
was known as "The London School."
 Introduction: Hans Eysenck and the study of intelligence 215

The basic philosophy and quantitative technology of this school of

psychology was characterized by its focus on the study of individual
differences, also known as differential psychology, with its central purpose of
discovering and objectively measuring all of the main mental and behavioral
traits and capacities that showed individual differences among human beings.
Also, it aimed to discover the causes of these measurable individual differ-
ences, causes attributable to differences in persons' heredity, biological struct-
ures, and life experiences. Differential psychology originated with the unique
and powerful influence of Sir Francis Galton (1822-1911) in British psycho-
logy. A scientific genius, Galton was the father of psychometrics, behavioral
genetics, and a number of basic concepts used in psychological statistics.
Besides his own original contributions to behavioral science, Galton also
promoted the biological and evolutionary thinking of Charles Darwin (who was
Gallon's half-cousin) and Herbert Spencer in psychology.
Historically, the founding fathers of the two main branches of psychology as
an empirical science, distinct from philosophy, were Wilhelm Wundt (1832-
1920) in Leipzig (experimental psychology) and Francis Galton in London
(differential psychology). The oldest, most established universities of
England—Oxford and Cambridge—followed mainly in the experimental
tradition established by Wundt, while psychology in the University of London
developed predominantly along Galtonian lines, even though, strangely
enough, the psychology department's preeminent head, Charles Edward
Spearman (1863-1945), was the only professor in Britain who received a Ph.D.
under Wundt.
But Galton was really the father of mental measurement. He devised
laboratory devices for the measurement of many psychological and anthro-
pometric traits and calculated correlations between these two classes of
variables. Although Galton never held any official academic position, he was
instrumental in the founding of both the departments of psychology and
genetics in the University of London, and he contributed various of his own
"brass instrument" inventions for measuring human characteristics to the
newly founded psychological laboratory in London's University College.
Galton's writings had considerable influence among the generation of
students at the turn of the century who were just embarking on careers in
academic psychology. The so-called "London School," to which Eysenck was
introduced as a student majoring in psychology in 1935, is itself the product of
nearly 50 years of domination of its Psychology Department by two of Britain's
most eminent psychologists and ardent exponents of Galton. Both men had
been strongly influenced by Galton's work and they promoted his ideas
throughout their careers. Spearman, while a student under Wundt, read
Galton's major works with much greater interest, he claimed, than he had
found in any of Wundt's work. While he was still a doctoral student in Leipzig,
two years before receiving his degree, he published (in The American Journal of
216 Intelligence

Psychology, 1904) a study that was most iwWundlian in its subject matter and
its methodology, but strikingly Galtonian. Titled General Intelligence, Objective-
ly Determined and Measured, it became widely recognized as a landmark in the
history of psychometrics and differential psychology. Not only was it a pivotal
contribution in its own right, but it laid out virtually all of the issues that, with
further development, became Spearman's greatest contributions to psychology,
including the discovery of g, the invention of factor analysis, and laying the
foundation of what is now known as classical test theory.
The second preeminent figure and Gallon exponent to dominate the
London School as Spearman's successor in 1932 was Professor Sir Cyril Hurt
(1883-1971). Burt had previously served for seven years as the professor of
educational psychology in London University. Just three years after Burt
assumed the Chair of Psychology, Eysenck enrolled in his department. Though
officially retired, Spearman remained in the department, and Eysenck came to
know the great man personally. I recall once hearing Eysenck contrast Spear-
man with Burt; he remarked that Spearman was not only a great psychologist,
but also a nice fellow (presumably being unlike Burt in this respect).
The then well-established tradition of the London School was the basis of
Eysenck's first view of psychology. And indeed, Burt's major research interest,
like Spearman's, was in the theory and measurement of human mental
abilities. It is noteworthy that Eysenck's first publication, in 1939 (a year before
he received the Ph.D.), was a review of L. L. Thurstone's Primary Mental
Abilities, in which Eysenck applied Burt's method of factor analysis to
Thurstone's correlation matrix of 56 diverse tests and showed the existence of a
large general factor, equivalent to Spearman's g, on which all of the tests were
loaded and which accounted for more of the total variance than any of the
primary factors identified by Thurstone's method of multiple factor analysis,
which rotated the several primary factors to simple structure, a procedure that,
of mathematical necessity, obliterated the general factor, creating the false
impression that a general factor did not exist in the matrix. The method of
factor analysis advocated by Burt (and used by Eysenck on Thurstone's
correlation matrix) allowed the emergence of a large general factor, while also
preserving each of Thurstone's primary factors, represented as first-order
factors independent of the second-order factor, g.
Given this beginning and these influences, in addition to Eysenck's early
recognized outstanding academic ability (Burt himself told me, in 1971, that he
thought Eysenck was his "most brilliant and most industrious" student), it
seemed inevitable that Eysenck was appointed Reader (1950) and then
Professor of Psychology (1955) in the University of London, and headed the
postgraduate Psychology Department in the Institute of Psychiatry. Nor is it
surprising that Eysenck has long since rightfully earned a reputation as the
leading contemporary exponent of the Gallon tradilion and Ihe London
School.
 Introduction: Hans Eysenck and the study of intelligence 217

Probably because research on intelligence was so completely dominated by

Burt during the early phase of Eysenck's career, and because during World
War II he was appointed psychologist in London's Mill Hill Emergency
Hospital, a psychiatric center in which the practice of clinical psychology
predominated, he did not consider doing research on intelligence until many
years later. Instead, he focused his research expertise in psychological
measurement and factor analysis on a much less developed field—the scienti-
fic study of personality. He envisaged a major program of research in this new
area similar to that conducted first by Spearman and later by Burt for the
subject of mental abilities. His empirical studies on personality soon issued
forth in several influential books, Dimensions of Personality (1947), The
Scientific Study of Personality (1952), and The Structure of Human Personality
(1953), and in countless journal articles. Research on the taxonomy, factor
analysis, and causes of personality differences, including the well-known
psychiatric syndromes, was the major effort of his newly organized Department
of Psychology at the Institute of Psychiatry, a setting in which research on these
topics was naturally most appropriate. The innovative methods advocated by
Eysenck fortunately were favored and encouraged by the imposing medical
head of the Institute of Psychiatry, Professor Sir Aubrey Lewis, one of Britain's
eminent psychiatrists.
Although the bulk of Eysenck's research contributions have been in the
personality field, he has also been remarkably prolific in his contributions to
the literature of human intelligence. There were chapters related to intel-
ligence in his first popular Penguin paperback, Uses and Abuses of Psychology
(1953), but his first theoretically important essay in this field did not appear
until 1967. (I will say more about it in my essay that follows.) Since then,
Eysenck has published about 50 articles, six books, and numerous book reviews
that all deal with human intelligence. Most of these items, up to 1985, are listed
in the references of my chapter on Eysenck's contributions to the study of
intelligence in Hans Eysenck: Consensus and Controversy edited by S. Modgil
and C. Modgil (1986). (The same volume contains a detailed and purposely
somewhat critical examination of Eysenck's theoretical approach to intel-
ligence by J. S. Carlson and K. F. Widaman.)
Since I have already written at length about Eysenck's theory of intelligence
in the Modgil and Modgil (1986) volume, I will here only characterize his
contributions in broad outline and leave it up to the other contributors in this
part to discuss the specifics. As I have pointed out, Eysenck's approach to
intelligence can be characterized in general by the terms objective, quanti-
tative, analytical, and biological.
The first empirical research on human intelligence that I am aware of in
Eysenck's career, done in collaboration with one of his colleagues at the
Institute, Desmond Furneaux, conceived of "splitting" Spearman's g into three
psychologically distinct components: mental speed, error checking, and
218 Intelligence

persistence. With a grant from the Nuffield Foundation, Furneaux developed

an individual test, which he called the Nufferno Test, based on number series
(which Spearman and others had found to be highly g-loaded). The testee was
closely observed while taking the test and the solution time for each single item
(each presented on a separate card) was individually measured. This permitted
separate scores to be derived for each of the components—speed, error
checking, and persistence (measured on the more difficult problems).
Furneaux proposed a theory to explain individual differences in the speed
score, based on the speed of a hypothetical neural scanning mechanism in the
brain. The other two components, however, appeared not to be cognitive
variables but really belong in the personality domain. What had actually been
"split" by Furneaux's method was not g per se, but the particular measure of g
obtained with the Nufferno test. This is an important point from a psycho-
metric standpoint, because the particular mechanics and item content of any
test of homogeneous item content always results in the test scores containing
other components of variance besides g. The Nufferno test separated out at
least two of these non-g factors as distinct scores. Cognitive tests with other
properties would contain other non-g factors. In a factor analysis of a great
many diverse tests, as both Spearman and Burt had shown, g could be
separated from all of the non-g group factors and test specificity. But no
systematic effort was ever made to demonstrate the relationship of the
Nufferno's measures of error checking and persistence to other g-loaded tests,
and the Nufferno test had little subsequent influence on intelligence theory or
research. What persisted as central in Eysenck's thinking, however, was the
hypothesis of individual differences in mental speed as the basis of g.
Similarly, Raymond B. CattelPs claim to have split Spearman's g into fluid
(Gf) and crystallized (Gc) components by means of hierarchical factor analysis
appears to be attributable to the limited variety of tests entered into the factor
analyses that show this split. Gf and Gc are themselves highly correlated in
most populations and if the hierarchical factor analysis is not truncated at the
level of second-order factors but a third-order factor is extracted, it turns out to
be Spearman's solitary g. Recent hierarchical factor analyses show that the
larger and the more diverse the battery of tests that is factor analyzed, the
greater the similarity between Cattell's second-order Gf factor and the third-
order g factor. In fact, when a large enough number of highly diverse tests are
included in the factor analysis, Spearman's g and Cattell's fluid Gf become one
and the same general factor, org. Gf disappears as a second-order factor, being
absorbed into, and identical with, the higher order g at the third stratum, or
apex, of the hierarchical factor structure. Gc, however, remains as a rather
minor second-order factor, loaded mainly in various tests of scholastic and
cultural achievements. It now seems a safe generalization that the g that
emerges from the factor analysis of complex psychometric tests, such as those
 Introduction: Hans Eysenck and the study of intelligence 219

used in the factor-analytic work of Spearman, Burt, and Cattell, has not yielded
to being "split" by any means available at the psychometric and factor analytic
level of analysis.
Eysenck realized that to pursue the analysis of g any further would require
that the cognitive measurements be obtained at a different, more elementary,
level of information processing than is possible for conventional psychometric
tests. The greater complexity of mental processes that conventional test items
call upon for solution (or for initially having been consolidated in long-term
memory) rendered conventional tests unamenable to further analysis beyond
the item level. Eysenck was one of the earliest to see the necessity for going
beyond the factor analysis of ordinary psychometric tests, and in 1967 he
spelled out his ideas for a new, more analytic, approach to intelligence research
("Intelligence assessment: A theoretical and experimental approach." British
Journal of Educational Psychology, 37, 81-98). In this innovative article, and
even more so in many subsequent works, Eysenck put forth compelling
arguments from the philosophy of science aimed at making sense (and
countering the prevalent nonsense) on the topic of intelligence, which, outside
the range of influence of the London School, was encumbered with dis-
gracefully unscientific and muddled misconceptions, some unfortunately still
blighting modern psychology textbooks. Eysenck's writings in this field showed
him, above all, to be a wonderfully clear thinker in a field that most needed
clear thinking. This was probably the chief source of his appeal to those
reading him for the first time. Whatever else may be said about Eysenck, his
core values, never ideological or dogmatic, have always embraced an un-
compromising empirical science approach to any and all psychological
phenomena. This does not insure always being right in every hypothesis or
every conclusion, of course, but it certainly does keep investigation on the right
track, and on that score Eysenck has pursued an unwavering course. His
explicit insistence on psychology as natural science is absolute. Given this
philosophy, inevitably the scientific truth will out.
Eysenck's suggested approach to research on intelligence was at first actually
more Galtonian than Spearmanian, both in its barkening back to Galton's idea
of mental speed as the basis of individual differences in general ability (which is
itself a Galtonian concept) and its advocacy of various forms of reaction time
(RT) as a tool for the measurement and analysis of individual differences. (A
then recent study by Roth [1964], published in Germany, had shown a
substantial correlation between RT and a conventional measure of g.)
RT in elementary cognitive processes soon became a lively area of research
in differential psychology and when the relationship between RT and
intelligence (as presaged by Galton but incapable of demonstration by his
relatively primitive methodology 100 years earlier) became clearly evident in
the burgeoning research literature on RT, Eysenck made the next logical
step—taking up the measurement of the one physiological variable already
220 Intelligence

shown to be correlated with IQ, the brain's electrical activity in response to an

external stimulus, measured as the averaged evoked potential, or AEP. The
discovery of a physiological basis of g was not only implicit in the work of
Gallon, but was explicit in Spearman's major work, The Abilities of Man (1927),
in which he expressed the hope that a neurological explanation of g would be
discovered—"whereby physiology will achieve the greatest of all its triumphs"
(p. 407).
But the technical means for direct physiological research on g, instead of
merely theoretical speculation about brain processes, only became available in
the 1960s. In the following decade, Eysenck began promoting such research in
his laboratory, measuring the AEP and nerve conduction velocity and
correlating these variables with IQ, and contributing many theoretical and
substantive articles and influential book chapters in this line, often in
collaboration with his students and colleagues (such as Elaine and Alan
Hendrickson and Paul Barrett) who had become expert in electrophysiological
methods. Indeed, it might be said that a major step toward Spearman's
expressed wish occurred when Eysenck and co-workers demonstrated that the
column of correlation coefficients showing the degree to which each of the 11
subtests of the Wechsler Adult Intelligence Scale (WAIS) is correlated with
the complexity of the waveform of the AEP was directly proportional to the
corresponding column of each of the 11 subtest's loadings on the g factor. The
rank-order correlation between the two columns was +0.95! Later that same
year (1985), and independently of Eysenck's report, a pioneer of AEP
research, E.W.P. Schafer, using the WAIS and the same method as described
above but based on a different AEP measure (the habituation of the amplitude
of the AEP), showed a rank-order correlation of +0.92, very similar to that
reported by Eysenck. In brief, there is an intimate relation between
Spearman's g as reflected in psychometric tests and the complexity and
amplitude of brain waves in response to external stimuli.
Thus, it appears clear to me that Eysenck has cast a greater influence than
any other contemporary psychologist in advancing what certainly must be
referred to henceforth as the Galton-Spearman-Eysenck school of research on
human mental ability.
Chapter 11
The psychometrics of intelligence

A. R. Jensen

1. INTRODUCTION
1.1 The specificity doctrine
My research interest in human mental abilities grew out of the viewpoint
espoused by Eysenck that I referred to in my introduction as the Galton-
Spearman-Eysenck school. In several of his publications, Eysenck has
contrasted this school of thought about the nature and measurement of
mental ability with the other major approach stemming from the work of
Alfred Binet (1857-1911).
At the behest of the Paris school system, Binet (in collaboration with
Theophile Simon) invented what was probably the first practical test of
intelligence. The Binet-Simon test was also the first mental test to be scaled in
mental-age units. It soon proved highly useful for the diagnosis of mental
retardation among children for whom schooling presented unusual difficulty.
This was a major achievement and is unquestionably a landmark in the history
of mental measurement. Binet was a distinguished experimental psychologist,
Simon a physician, but neither one was a psychometrician in the technical
sense of that term as we understand it today. In fact, the branch of psychology
known as psychometrics had not even come into existence when Binet's test
was published in 1905. Because the basic principles of psychometrics still
awaited Spearman's formulation in terms of what is now known as classical test
theory, Binet and his immediate followers could not possibly have conceived or
described the "intelligence" measured by Binet's test except in terms of the
test's superficial features, such as its specific item content (vocabulary,
counting, form board, paper folding, esthetic judgment, arithmetic operations,
matching forms, etc.), and the inferred mental faculties that the test items
supposedly called upon (memory, discrimination, comparison, reasoning,
judgment, etc.). As Binet's test was considered a measure of intelligence, and
its overall scores (or the derived mental-age) accorded quite well with teachers'
subjective estimates of their pupils' intelligence as judged from their classroom
behavior and scholastic performance, psychologists naturally defined intelli-
gence as the sum of all the kinds of knowledge and skills visibly recognized in
222 Intelligence

the items of the Binet test (or of other tests modeled after it). At a somewhat
higher level of abstraction, intelligence was described in terms of all the various
faculties that were surmised to be called for by the particular cognitive
demands made by each of the different types of test items.
I have termed this prevailing conception of what is measured by intelligence
tests the "specificity doctrine." From a strictly psychometric standpoint, it is
easy to prove that the specificity doctrine is absolutely wrong, despite the fact
that it is the prevailing notion among many clinical and applied psychologists—
the very psychologists who use tests the most.

1.2 The fallacy of the specificity doctrine

Spearman was the first trenchant critic of the Binet approach and of the
wrongly conceived specificity doctrine, arguing that its conception of
intelligence and its method for measuring it were based on an arbitrarily
chosen hotchpotch of items that include various kinds of knowledge and
cognitive skills. There is nothing magical about the particular collection of the
items that compose the Binet test. Any other equally diverse collection of items
that tap many different bits of knowledge and skills would do just as well.
But why should this be so? Given the minimum requirement that the items
range widely enough in their level of difficulty to produce reliable individual
differences in the total score, why is the specific item content of the Binet test
(or any other test) of so little importance? Spearman's complaint with Binet's
signal contribution was not so much with the test itself as with the mis-
conception about what it measures, a misconception, incidentally, that has
endured among many psychologists up to the present day. Spearman's key
discovery of what the Binet test (and all other tests of its type) essentially
measures was completely unrealized by Binet. It is my impression that it is no
better understood by many present-day psychologists, not to mention the
popular media and the general public.

1.3 The signal and the noise

To make a long story short, Spearman was the first person to recognize that the
total score on a mental ability test does not measure the ability to do this item
or that, such as being able to define a particular word, know a particular fact,
comprehend a given sentence, recall a string of digits, solve a particular puzzle,
or copy a particular form. The observed individual differences in overall test
scores that we represent as the true-score variance in some defined population
(or a sample thereof) does not reflect individual differences in the specific
abilities needed to perform correctly on this, that, or other items of a test. We
can easily prove this by looking at the correlation between persons' scoring
"right" or "wrong" on any given item and that item's correlation with any other
 The psychometrics of intelligence 223

item in the test. In the best IQ tests, the average interitem correlation is typically
between +.10 and +.15, and the correlation between any single item and the total
score on a test composed of many diverse items is typically between +.20 and
+.30. In other words, whatever it is that is measured by the test as a whole is
measured only very slightly by each of the test items. The total variance of each
item contains only a faint "signal," which reflects a source of variance (i.e., a
factor) that is common to each and every one of the disparate items that make up
the test. The vast bulk of the variance on any given item, however, is just "noise."
What we can see with the naked eye by examining various test items is only this
noise, that is, the specific bits of knowledge or skill called upon by a particular
item. For any given item, the signal to noise ratio is typically about 1 to 99. But by
aggregating the scores on each of a large number of diverse items to obtain a total
score, the signal to noise ratio for the total score is typically greater than 9 to 1.
The reliability of a test, that is, the percentage of its true-score variance, is about
90% of the total raw-score variance for most professionally developed
standardized mental tests. The remaining 10% of the variance, or the noise, is
the sum of the item variances. By examining each of the items to determine what
the test measures, all that one really sees is the noise (or error component) of the
total test scores. In other words, what we measure as individual differences, or
variance, in total scores mostly consists not of item variances, but of twice the sum
of all the covariances among items. In the total variance of test scores, these two
sources of variance are generally in the ratio of 1 to 9. Unlike test items,
covariances are not "things," or bits of knowledge, or specific skills. One can see
the item content (the noise) by inspecting the items. But one cannot see the item
covariances (the signal), which can only be determined by analytical calculations.
The item covariances originate from the aggregation of the small invisible
signal reflected by each item. (The sum of the item variances, or noise,
constitutes the test's error variance.) The test's true-score variance is solely
attributable to the fact that the diverse items are all positively correlated with
one another, however slightly, thereby producing a total of n(n—1) covariance
terms (where n is the number of items in the test). The sum of all the item
covariances constitutes the test's true-score variance. But the important point I
wish to make here is that the main source of the variance of interest when we
use a test does not reside in the test items per se, but reflects whatever it is that
causes items to be positively correlated with one another. The positive correl-
ation between all cognitive test items is a given, an inexorable fact of nature.
The all-positive interitem correlation matrix is not an artifact of test
construction or item selection, as some test critics mistakenly believe. In fact,
it is empirically impossible to devise a cognitive test that has nonzero variance
and in which there are statistically significant negative interitem correlations.
An imaginary test for which the average interitem correlation is either negative
or zero is the psychometric equivalent of a perpetual motion machine.
224 Intelligence

1.4 Distilling psychometric g

The variance attributable to the sum of the factors derived from a factor
analysis constitutes the common factor variance, of which the largest compo-
nent is usually the g factor. (Other common factors typically found in mental
ability tests are verbal, spatial, numerical, and memory.) The g factor, emerging
as it does from correlations among elements rather than from the addition of
elements, is not a compound or conglomerate, but rather is more aptly likened
to a distillate.
Although the g factor is typically the largest component of the common
factor variance, it is the most "invisible." It is the only "factor of the mind" that
cannot possibly be described in terms of any particular kind of knowledge or
skill, or any other characteristics of psychometric tests. The fact that psycho-
metric g is highly heritable and has many physical and brain correlates means
that it is not a property of the tests per se. Rather, g is a property of the brain
that is reflected in observed individual differences in the many types of
behavior commonly referred to as "cognitive ability" or "intelligence."
Research on the explanation of g, therefore, must necessarily extend beyond
psychology and psychometrics. It is essentially a problem for brain
neurophysiology.

1.5 The Galton-Spearman-Eysenck school

In order to focus theory and research in this domain, Eysenck, probably more
than anyone else, has emphasized the importance of D. O. Hebb's classic
distinction between three different meanings of "intelligence," which are
labeled Intelligence A, Intelligence B, and Intelligence C. Failure to observe
these crucial distinctions only obscures theoretical discussion and creates
spurious arguments. Intelligence A refers to the biological (i.e. genetic and
neurophysiological) basis of observed individual differences in Intelligence B
and Intelligence C. Intelligence B is any form of gross observable behavior that
involves cognitive abilities as they are manifested in "real life" circumstances—
in learning, problem solving, memory, general knowledge, verbal facility,
quantitative reasoning, levels of mastery of vocational skills, educational and
occupational level, income, social adeptness, intellectual interests and achieve-
ments, and so on. Intelligence C is the cognitive ability measured by psycho-
metric tests, such as IQ.
These three conceptions of intelligence are of course not independent, and
all are worthy of study in their own right. But Intelligences A and C are far
more amenable to exact scientific study than Intelligence B. The class of
variables that fall under Intelligence B is so unbounded, and the causes of
individual variation at this broad-scope level of observation involve such a
multiplicity of social, cultural, experiential, motivational, and specific con-
textual factors as to frustrate attempts to arrive at a scientifically satisfying
 The psychometrics of intelligence 225

theoretical formulation. About all we can do with Intelligence B is to

determine the degree of correlation some of its quantifiable aspects have with
both Intelligences A and C, and possibly, by using the techniques of path
analysis, we can test competing hypotheses concerning the causal relationships
between Intelligences B and C.
Intelligence C, such as IQ, is itself problematic in that it reflects to some
extent the particular item composition of the test, more or less, depending on
the number and diversity of the items, which is technically referred to as the
adequacy of psychometric sampling. Differences in psychometric sampling
account for why various IQ tests are not perfectly correlated, although the
correlations even among standard IQ tests that differ markedly in item
composition is quite high, averaging about + .80 (which goes up to about + .90
after correction for attenuation [unreliability]). Because all cognitive tests,
without exception, are positively correlated with one another and therefore,
when factor analyzed together, yield a general factor, g, it makes sense to use g
as the criterion of the adequacy of any given test as a measure of general
mental ability. Thus a measure of Intelligence C, to the extent that it is
correlated with g, is the best psychometric marker available for discovering
which biological variables are involved in Intelligence A. In fact, we now know
empirically that the larger the test's g loading (Intelligence C), the larger its
correlation with biological variables (Intelligence A). In other words, the
process of extracting the g factor from a large and diverse battery of tests
screens out, so to speak, much of the variance in conventional intelligence test
scores that is unrelated to Intelligence A. The g factor is probably closer to the
biological substrate, Intelligence A, than any other measure we can derive from
conventional psychometric tests. A figure often used by Eysenck represents the
variance as the area of a circle and shows the causal influences (arrows) of
different kinds of variables on Intelligences A, C, and B, which in Figure 11.1
are labeled as biological, psychometric, and social intelligence, respectively.
The Galton-Spearman-Eysenck school focuses primarily on Intelligences A
and C. Its research program in this field is aimed first at discovering the
relationship of psychometric g to biological variables, and second at ultimately
explaining the causal nature of the relationship in terms of neurophysiological
mechanisms. A number of possible approaches are available: reaction time
(RT) and inspection time (IT) measures of the speed of information
processing in elementary cognitive tasks, and physiological measures such as
the average evoked potential (AEP), neural conduction velocity, the brain's
glucose metabolic rate (GMR) during problem solving, and brain size
estimated from head measurements or measured directly by magnetic
resonance imaging (MRI). Each of these is worth pursuing, because the
different kinds of data they provide, when viewed in conjunction, may serve as
triangulation points for hypothesizing the nature of the information processes
and their biological mechanisms that are reflected by the g factor derived from
226 Intelligence
Socioeconomic
Cultural Family status Motivation
Genetics factors upbringing Experience Nutrition
Physiology
Health Cultural
factors

Drinking
Personality habits

Coping
Education
Education Socioeconomic strategies
status Mental Family
disorders background

BIOLOGICAL PSYCHOMETRIC
INTELLIGENCE INTELLIGENCE

Figure 11.1. Eysenck's representation of three different conceptions of "intelligence." Biological

intelligence is measured by the electroencephalogram (EEG), the average evoked potential (AEP),
nerve conduction velocity (CNV), the galvanic skin response (GSR), and reaction time (RT).

conventional psychometric tests. For the most recent synthesis of Eysenck's

whole philosophy about the nature and aims of research on intelligence,
readers are referred to Eysenck (1994).

2. REACTION TIME AND PSYCHOMETRIC g

2.1 Introduction
Until 1964,1 had never given any thought to reaction time (RT) in relation to
intelligence. I had learned at some time in the past, probably as an under-
graduate psychology student, that way back in "ancient history" Gallon had
hypothesized a correlation between RT and mental ability of the type pre-
viously described as Intelligence B (since there were no IQ tests at that time). I
also knew that it was common knowledge among psychologists that Gallon's
idea on this point had long since been completely discredited empirically and
that even to suggest that it might possibly still have some merit was a sign of
ignorance or naivete.
In 1964-1965 I spent the academic year of my first sabbatical leave from
Berkeley in Eysenck's department at the Institute of Psychiatry, where seven
years earlier I had spent two years on a postdoctoral fellowship from the
National Institute of Mental Health. I recall that one day in 1964, during the
mid-morning coffee break in the old Maudsley cafeteria, Eysenck joined
several of us at our table and told us in considerable detail and with evident
enthusiasm about an article he had recently read in a German psychology
journal (Roth, 1964). It was a study of RT in which the subject, seated in front
 The psychometrics of intelligence 227

of a display panel, was instructed to turn off a light as quickly as possible after
its appearance on the panel by touching a button adjacent to the light. The
light that went on (and that was to be turned off by the subject) was presented
on the panel under each of several different conditions, referred to as set-size:
the light appeared either alone (set-size 1), or appeared among sets of 2,4, or 8
lights. In each set-size, any one of the alternative lights would go "on" at
random. Each subject was given a large number of trials under each set-
size and the subject's RT on each trial was measured in milliseconds (ms). (RT
is the interval between the light's going "on" and the subject's touching the
button that turns it "off.") When the RTs for each set-size were averaged over
trials and over subjects, the phenomenon known as Hick's law was clearly
evident. If n is set-size (i.e.,the number of alternative possibilities for which of
the n lights would go "on"), Hick's law states that the increment in RT for each
unit of increase in set-size is equal to the binary logarithm of the set-size, or
ART = k + Iog2« (where k is the RT for n = 1). In information theory, the
unit of information, termed a bit (for binary digit), is measured as the binary
logarithm (Iog2) of the number of alternatives; a bit is the amount of
information needed to reduce uncertainty by one-half. Thus, Hick's law
expresses the linear relationship between amount of information and RT.
But what Eysenck found to be exciting about Roth's study was not that it
confirmed Hick's law (about which there was little doubt, in any case), but that
it showed rather marked individual differences in the slope of the Hick
function (i.e., the linear regression of RT on bits) and, most important, these
individual differences in slope had a significant negative correlation with IQ.
Gallon's original conjecture, via Hick's law, appears to have been sub-
stantiated. It made perfect sense theoretically that, if intelligence were
conceived of as the speed or efficiency of information processing, the rate of
increase in RT as a function of bits should be greater for persons of low IQ
than for persons of higher IQ. It was also of interest to see that here was an
exceedingly simple task that bore absolutely no resemblance to any IQ test and
yet the RT parameters (median RT and the slope of RT/bits) derived from it
were correlated with IQ.
Right then and there, for some reason that I cannot fathom, I felt that this
was perhaps the single most interesting finding I had come across in my total
acquaintance with psychology up to that time. But I did nothing about it. In
retrospect, I think what I should have done was to drop everything else and
immediately go to work on this Hick paradigm and its relation to IQ. But when
I returned to Berkeley I was committed to completing other work in progress
and I had research grants for experiments on the psychology of human
learning. Then I was invited to spend a year at the Center for Advanced Study
in the Behavioral Sciences, where laboratory work was not possible. And so it
went, year after year. Although I went on thinking about the Roth experiment,
of and on, year after year, I found no way to work it into my agenda.
228 Intelligence

Then, in 1967, Eysenck sent me a reprint (Eysenck, 1967) of what, for me,
was one of his most important articles. In it, he put forth a theory of
intelligence that was essentially biological and Galtonian, hypothesizing the
speed of information processing as the basis of individual differences in
performance on cognitive tasks. He also explained how Roth's 1964 study of
the correlation between RT and IQ in the Hick paradigm jibed with Gallon's
ideas and brought research on individual variation in mental ability closer to its
biological basis than was possible with conventional psychometric tests. My
long-standing latent interest in this approach, then having been energized by
Eysenck's stimulating article, I applied for a grant to construct the kind of RT
apparatus needed to pursue this line of investigation.

2.2 The Hick paradigm

Figure 11.2 shows the subject's response console for the present model of my
apparatus (in the first model, the lights and their closely adjacent pushbuttons
were separate; in the present model, the pushbuttons themselves would light
up, thereby maximizing the stimulus-response compatibility of the RT task).
The whole sequence of practice trials and stimulus presentations is run by
computer, which also records response times (in ms) and errors on each trial.
The one feature that differs importantly from the procedure used by Roth is
the use of a "home" button, which makes it possible to separate RT from
movement time (MT). RT is the time interval between the onset of the
reaction stimulus (i.e., one of the buttons lighting up) and the subject's
releasing the home button. MT is the interval between the subject's releasing
the home button and pressing the lighted button to turn it off. Typically 15-30
trials are given for each set-size. From these data, five chronometric scores are
derived for each subject: (1) median RT, (2) median MT, (3) the standard
deviation of RT over trials (RTSD), (4) the standard deviation of MT over
trials (MTSD), and (5) the slope of RT across set-sizes expressed as bits.
Figure 11.3 shows the typical results for a group of subjects.
The characteristic features seen in every study (except those based on
severely retarded subjects with IQs below 50) are (1) the significant linear
slope of RT as a function of bits, (2) the near-zero slope of MT as a function of
bits, and (3) that RT is much greater than MT even at 0 bits.
My program of research on the relationship between RT and psychometric g
has extended over a period of more than two decades, and it would be
impossible here even to summarize all of the findings within the allotted limits
of this chapter. Most of these, however, are summarized in considerable detail
elsewhere, along with references to most of the original studies (Jensen, 1982,
1987a, 1987b, 1993a). Here I will simply note some of the main findings,
explain some of the main controversies that have since been more or less
resolved, and note some of the open questions that await further research.
 The psychometrics of intelligence 229

Figure 11.2. The subject's response console of Jensen's RT-MT apparatus. The panel is 13 in. x
17 in., painted flat black, and tilted at a 30 angle. At the lower center is the home button (black, 1
in. diameter), which the subject depresses with the index finger while waiting for the reaction
stimulus. The semicircle of eight small circles represents translucent pushbuttons (green, 0.5 in.
diameter, each at a distance of 6 in. from the home button); each button can be lit independently.
Touching a lighted button turns off the light. Various plates can be placed over the console to cover
some of the buttons, leaving either 1,2,4, or all 8 buttons exposed to view, making for four different
ECTs, each with a different number of equally likely response alternatives. The binary logarithms
of 1, 2, 4, and 8 exposed buttons are equivalent to 0,1, 2, and 3 bits of information, respectively. A
trial begins with the subject depressing the home button; 1 s later a preparatory stimulus ("beep")
of 1 s duration occurs; then, after a 1-4 s random interval, one of the buttons lights up, whereupon
the subject's index finger leaves the home button and touches the underlighted button. RT is the
interval between a light-button going "on" and the subject's lifting the index finger from the home
button; MT is the interval between releasing the home button and touching the underlighted
button. In each trial only one of the buttons lights up, entirely at random from trial-to-trial.

2.3 Main findings

The basic generalization that follows from the results of this research is that
Gallon's hypothesis that speed of reaction and general intelligence are
intimately related is amply substantiated. Innumerable studies, from my own
230 Intelligence

500r-

MT

Bits
Figure 11J. The median RT and MT obtained on the RT-MT apparatus (Figure 11.2) averaged
over more than 1500 individuals. Note the significant positive slope of RT (RT = 336 + 34 BIT, r
= .998) demonstrating Hick's law, which predicts a linear relationship of RT to the amount of
information measured in BITs. In marked contrast is the nonsignificant slope of MT (MT = 245 +
4.3 BIT, r = .641) (data from Jensen, 1987, Tables 3 and 7).

laboratory and from many others around the world, have shown correlations,
mostly in the range of .30 to .50, between response times and scores on
untuned g-loaded tests, such as IQ. There is no longer the least doubt about
the fact of correlation between RT and g. It shows up not only in the Hick
paradigm (described above), but in every elementary cognitive task (ECT) that
uses RT as the measure of proficiency. The correlations are largest for ECTs in
which the RT is less than 1 s for adults or less than 2 s for elementary school
children. Within this narrow range, the degree of correlation between RT and
IQ is related to the complexity of the cognitive demands made by the ECT
used to elicit RT. If the task is too complex and results in longer RTs, other
factors besides speed of processing intervene to attenuate the correlation
between RT and g.
 The psychometrics of intelligence 231

Moreover, we have found that RT is more related to g than to any other

factor independent of g that can be extracted from a battery of psychometric
tests. RT correlates with various psychometric tests to the degree that the tests
are g loaded.
RT in a variety of ECTs shows very large and highly significant differences
between criterion groups that differ in mental ability not only as measured by
psychometric tests, but as recognized by common sense, such as:

1. Between the institutionalized retarded or persons in sheltered workshops

and the general run of "normal" persons.
2. Between precocious children who are succeeding in college at 12-14 years
of age as compared with their age-mates who are in the usual school
grades for their age.
3. Between students in a selective university and students of the same age in a
much less selective vocational college.
4. Between young adults (ages 18-22) and much older adults (ages 65-80) of
similar socioeconomic background and level of education (the older adults
have slower RT, especially on the ECTs that are most highly correlated
with g).

RT and MT do not measure the same thing. They are correlated only about
.30, and MT is almost completely unrelated to experimentally manipulated
variations in the complexity of any particular ECT, whereas the manipulation
of complexity has marked effects on RT as well as its degree of correlation with
g. Moreover, in a factor analysis of RT and MT measures from a large number
of quite different ECTs, RT and MT very distinctly come out on different
factors, and MT has no significant loadings on g or on any other factors with
salient loadings on psychometric tests.
Intraindividual variability in RT, measured as the standard deviation of the
individual's RTs over a specified number of trials (hence labeled RTSD), has
lower reliability than the individual's mean or median RT, and yet RTSD is
generally more highly correlated (negatively) with g. Higher IQ persons
maintain more consistent RTs from trial to trial than persons of lower IQ. This
shows up in the degree of skew in the distribution of RTs for a given individual.
In a large number of trials, low IQ and high IQ persons differ relatively little in
their fastest RTs, but they differ markedly in their slowest RTs—low IQ
persons produce many more slow RTs, which makes their RT distribution
more highly skewed. Median RT and RTSD are highly correlated, even when
based on independent sets of RT data, but each measure is to some degree
independently correlated with g, RTSD slightly more so than median RT.
Eysenck, I believe, was the first to suggest that RTSD reflects neural noise in
the brain, which impedes the transmission of information, which is also re-
232 Intelligence

fleeted in longer RT. However, the independent correlation of median RT and

RTSD with IQ indicates that RT is not entirely derivative from RTSD, but that
both aspects of RT—speed and consistency—are independently related to IQ.

2.4 Main controversies and possible solutions

In our studies we have almost always used nonspeeded or untimed tests
(usually Raven's Matrices) to measure g, instructing subjects to attempt every
item and to take as much time as they needed. Subjects are usually tested alone
in a quiet room, so there is no chance of their being paced by observing other
individuals who may complete the test sooner. This is important in order to
rule out speediness of test taking as the common factor that might account for
the correlation between RT and g. We have established beyond any possible
doubt that the RT-g correlation is not a result of the speed factor often found
in psychometric tests. The amount of time subjects take to complete a difficult
cognitive test and then- RTs on an ECT have a near-zero correlation.
Moreover, RT is more highly correlated with scores on untimed tests than with
scores on highly speeded tests. At one time, a number of psychologists thought
we were merely rediscovering with our ECTs and measures of RT the clerical
speed factor that had long since been identified by factor analyses of test
batteries that included highly speeded tests. But the very tests that best
measure the psychometric speed factor have the lowest correlations with RT
and also with g. In brief, RT and RTSD certainly do not measure speediness of
test taking.
How, then, can we explain why RT on ECTs that are so simple that all
subjects can perform them with very few or no errors are substantially
correlated with performance on untimed tests that involve complex reasoning
and that measure individual differences in terms of the number of correct
answers, which reflects the level of item complexity and difficulty at which the
subject can successfully perform?
At least one part of the answer involves what cognitive theorists refer to as
the capacity of working memory, that is, the amount of information that can be
retained and manipulated in conscious awareness before any information is
lost through interference from new input or decay of memory traces. The
importance of processing speed in the operation of working memory stems
directly from the capacity limitation and the rapid decay of information in
short-term memory (STM). The limited capacity of working memory severely
restricts the number of operations that can be performed at any one time on
the information that enters the system from external stimuli or from retrieval
of information stored in primary memory or in long-term memory (LTM).
Quickness of mental operations is advantageous because more operations per
unit of time can be executed without overloading the system. Also, because
there is rapid decay of stimulus traces in the sensory buffers and of information
 The psychometrics of intelligence 233

in STM, there is an advantage to speediness of any operations that must be

executed on the information while it is still available. To compensate for
limited capacity and rapid decay of incoming information, the individual
resorts to rehearsal and storage of information into LTM, which has a
relatively unlimited capacity. But the process of storing information in LTM
itself uses up channel space, so there is a "trade-off' between the storage and
the processing of incoming information. The more complex the information
and the more operations that are required on it, the more time that is
necessary, and consequently the greater is the advantage of speediness in all
the elementary processes involved. Loss of information due to overload
interference and decay of information that was inadequately encoded or
rehearsed for storage and retrieval from LTM results in a failure to grasp all of
the essential relationships among the elements of a complex problem needed
for its solution. Speediness of information processing, therefore, should be
increasingly related to success in dealing with cognitive tasks to the extent that
their information load strains the individual's limited capacity. The extreme
simplicity of the ECTs, for which RT is therefore the only reliable source of
individual differences, permits us to measure the speed of information
processing when the capacity of working memory is not threatened by the
complexity of the task. Increasing the complexity of the ECT, as in going from
1 bit to 3 bits in the Hick paradigm, increases the RT-g correlation. But when
the task complexity is so great as to exceed the capacity of many subjects'
working memory, then the number of erroneous responses, rather than RT,
becomes the stronger correlate of g.
It is noteworthy that RT correlates only with g and not with any other
psychometric factors independent of g, even when the reaction stimuli of the
ECT are specifically designed to be either spatial, or verbal, or numerical and
the psychometric tests with which the RTs are correlated are either spatial,
verbal, or numerical. Statistically remove the general factor from the three
types of psychometric tests and their correlation with the verbal, spatial, or
numerical RTs is virtually zero.
Thus RT is a rather ideal tool for experimentally studying the task variables
that cause a given task to be more or less g loaded. RT is highly sensitive to
rather subtle experimental manipulations of the quantity of the information
load of the task, which can be varied experimentally without in the least
altering the type of information content or any of the stimulus or response
aspects of the task. Increasing the task demands (resulting in RTs of not more
than about 1 s) is found to increase the g loading of the RT parameters. In the
Hick paradigm, for example, one-bit increments in the reaction stimulus cause,
on average, only about 30 ms. increments in RT. The correlation of RT with IQ
increases linearly from about —.20 for 0 bits of information to about —.30 for 3
bits. These are small but significant correlations, and their linear slope as a
function of bits is also significant.
234 Intelligence

2.5 Open questions

The original experiment by Roth (1964) based on the Hick paradigm reported
a significant correlation (—.39) between IQ and the slope of RT as a function
of bits. This was a theoretically important finding, which meant that as the
information processing demand of the task was increased, the advantage of
higher IQ increased, as reflected by the higher IQ subjects' relatively faster RT.
The majority of later studies, however, failed to replicate the relatively large
correlation between IQ and RT slope originally reported by Roth, and this
aspect of the RT-IQ relationship was more or less dismissed as if it had been
totally discredited. True, the N-weighted mean correlation based on 32 inde-
pendent studies was only —.165, though it is significant at p < .001. But the
test-retest reliability of the slope measure (determined in six studies) is only
.39. When corrected for attenuation, the RT slope correlates with IQ about
—.26. For comparison, in the same 32 studies, the disattenuated mean
correlation between IQ median RT is — .24 (p < .001), and the mean dis-
attenuated correlation between IQ and RTSD (i.e., intraindividual variability
of RT) is —.34 (p < .001). Thus when test-retest reliability is taken into ac-
count, the slope parameter shows, on average, about the same correlation with
IQ as the median RT, and RTSD has the largest correlation with IQ.
Groups that differ in their average IQ show sizable mean differences in the
Hick slope parameter, always in the theoretically predicted direction. For
example, school students in regular classes and students in academically
"gifted" classes differ .50 to .70 of a standard deviation (SD) in mean slope;
university students and vocational college students differ about .50 SD; and
black and white vocational students differ 0.34 SD (all of these differences
significant beyond the .01 level).
Thus the Hick slope parameter accords with the theoretical implications first
noted in Roth's study and elaborated upon in Eysenck's (1967) article. The
measure of RT slope, like median RT, appears to be correlated only with theg
component of a test's variance. When each of the 12 subtests of the Wechsler
IQ battery was correlated with slope in the Hick paradigm, there was a rank-
order correlation of —.83 (p < .01) between each of the subtests' g loadings and
the degree to which each subtest is correlated with slope. That is, a test's
correlation with slope proved a highly valid predictor of the test's g loading. If
the three parameters of RT—median RT, slope, and RTSD—can be inter-
preted as measures of the speed and efficiency of information processing, then
we may conclude that the speed and efficiency of information processing is at
least a part of g. When RT parameters from a number of different ECTs are
combined (either by multiple regression or by simple addition of their unit-
weighted z scores), the RT-IQ correlations approach -.60.
 The psychometrics of intelligence 235

2.6 The odd-man-out discrimination task

To increase the information processing demands of the RT task, using the
same RT-MT apparatus shown in Figure 11.2, Frearson and Eysenck (1986)
cleverly modified the procedure to create an odd-man-out discrimination task.
All eight of the light-button alternatives are in view on the subject's response
console. On each trial, three of the eight lights go "on" simultaneously; two of
the lights are always closer together than the third, which is the odd-man-out.
For example, if we imagine that the light-buttons are numbered from 1 to 8,
the odd-man-out pattern would be such as 1, 2, 4; or 2, 4, 7, and so on. (With 8
light-buttons there are 44 possible odd-man-out patterns.) The subject is
instructed to respond as quickly as possible to touch the odd-man-out button,
which instantly turns off all of the lights. Again, RT is the interval between the
three lights going on and the subject's releasing the home button. While the 3-
bit condition of the Hick yielded RT-IQ correlations of about —.30 in samples
of university students, the odd-man-out procedure resulted in RT-IQ correl-
ations of about -.60, which approximates the average correlation between the
individual subtests of the Wechsler scale and the Full Scale IQ, and is about
the size of the correlation between the Wechsler IQ and the Raven IQ in our
student population.
Using the odd-man-out procedure along with the Hick procedures (all using
the same response console), we tested over 800 white and black school children
in grades 4-6 to test a hypothesis of Spearman's using RT-MT measures rather
than ordinary psychometric tests. Spearman (1927) had suggested that the size
of the standardized mean black-white difference on various psychometric tests
is directly related to the test's g loading. Spearman's conjecture had already
been strongly borne out in 12 studies based on conventional tests (Jensen,
1985). The question was whether it would be borne out using RT in tasks that
varied in information load (and hence in g) but contained no specific
informational content. The tasks were so easy that the most difficult task in the
battery (the odd-man-out task) could be performed with 100% accuracy by 4th
to 6th graders with RTs of less than 1 s (the average being about 700 ms) The
g-loadings of the various RT and MT measures derived from the Hick and odd-
man-out paradigms were indeed correlated with the standardized mean black-
white differences on these chronometric measures in accord with Spearman's
hypothesis and the correlation was even higher than for conventional psycho-
metric tests. The correlation between the g loadings of the chronometric
measures and the mean black-white differences on those measures was .80,
p < .01 (Jensen, 1993b).

2.7 The non-% RT factor

For a long time, researchers thought there was a virtually inexorable correl-
ation ceiling, at about -.35, between RT and IQ. Research with the Hick
236 Intelligence

paradigm produced results that were generally consistent with this rather low
correlational ceiling, even when the RT measures were corrected for
attenuation or were made highly reliable by averaging RTs over a very large
number of trials. (I found that increasing the number of RT trials by any given
amount yields reliability coefficients that accord perfectly with the reliability
predicted by the Spearman-Brown prophecy formula.) The ceiling, therefore,
cannot be blamed on reliability. Increasing the information processing
demands of the task, as in the odd-man-out procedure, surely breaks through
the —.35 ceiling, but the odd-man RT has its own correlation ceiling, between
—.5 and —.6. As already noted, if the processing demands result hi RTs greater
than about 1 s (in university students), the RT-IQ correlation markedly
shrinks.
I hypothesized that the RT measured by any one procedure is much like a
highly homogeneous psychometric test, that is, one in which all of the items are
so equivalent as to be almost identical. We know that such an extremely
homogeneous psychometric test, whatever its content and however reliable its
scores, has a relatively low correlation with psychometric g. The most g-loaded
tests are those that have quite heterogeneous items in terms of their types of
cognitive demands. The Wechsler Full Scale IQ, for example, is much more g
loaded than any one of its relatively homogeneous subtests. The items of the
highly g loaded Raven Matrices superficially appear to be highly homogeneous,
but actually the Raven items demand many different types of problem solving.
Therefore, according to my hypothesis, it should be possible to obtain much
higher correlations if we combined the RTs from a number of quite different
ECTs, each of which was simple enough to minimize erroneous responses and
elicit RTs in the range below 1 s. As in ordinary psychometric tests, the
specificity of each ECT should average out, allowing the emergence of the
common factor in the RTs from each of the different ECTs, namely g. So we
used a battery of different ECTs, each of which theoretically tapped a different
information process (indicated here in parentheses): the Hick (stimulus
apprehension and choice), the odd-man-out (discrimination), the Neisser
paradigm (speed of visual scanning), the Saul Sternberg paradigm (speed of
scanning STM), the Posner paradigm (speed of accessing information in
LTM), the semantic verification test (speed of matching symbols with mean-
ings), dual tasks (divided attention between two tasks, thereby straining
working memory capacity), and inspection time (speed of making a simple
visual discrimination). Indeed, the simple summation of the RTs and RTSDs
obtained from all of these paradigms resulted in a correlation with
psychometric g slightly greater than .60 and approaching .70 after corrections
for attenuation and restriction of range in the college population. The
correlations are scarcely larger between different standard psychometric tests,
such as the Wechsler, the Raven, the Terman Concept Mastery Test, and the
Multidimensional Aptitude Battery.
 The psychometrics of intelligence 237

But a hierarchical factor analysis of the correlation matrix containing a

number of conventional psychometric tests along with RT measures derived
from various ECTs reveals why there is an inexorable ceiling to their
correlation with psychometric g, regardless of how many different RT
measures we may combine in a single score. The reason is that the variance
of each RT measure based on a different paradigm does not consist only of g
plus the specificity of each paradigm. There is another quite large factor
besides g common to all of the various RT measures, which can be called a
non-g RT factor (Jensen, 1994). (There is also variance that is specific to each
RT paradigm.) In other words, RT tasks all measure g to some extent, but they
also measure an RT factor that is unrelated to any factors measured by
conventional psychometric tests. The total true-score variance of RT is divided
between g, a non-g RT factor, and the specificity of the particular ECT. The g
and non-g components of RT vary with the complexity of the ECT, the g
component being larger in the more complex tasks. But the ubiquitous
presence of the substantial and apparently noncognitive RT factor rather
severely limits the practical usefulness of any ECT, or even the combination of
several ECTs, as an alternative method for measuring the same g factor that we
can measure quite accurately and efficiently with a standard psychometric test.
The noncognitive RT factor, which seems to reflect individual differences in a
purely perceptual-motor speed or coordination ability, may be of interest in its
own right, and it is presently being researched by personnel psychologists in the
Air Force for its possible predictive validity in the selection of recruits for pilot
training.

2.8 Top-down, bottom-up and physiological explanations

Hardly a month goes by without some new ECT for measuring RT appearing
in the psychological literature (particularly in the journals Intelligence and
Personality and Individual Differences). In nearly every study there is found a
significant correlation between RT and psychometric g, and each such study
usually throws some light on the experimental variables that affect this
correlation. What is still unclear is the precise basis of the correlation between
RT and the g derived from nonspeeded psychometric tests—whether it is a
matter of individual differences in RT being influenced by whatever higher-
level cognitive processes are possessed by high-g persons (who are high-g for
reasons having no causal relation to RT), or whether RT reflects differences in
the speed and efficiency of the basic neural processes that cause differences in
psychometric g. These two alternative possibilities are known as the "top-
down" versus the "bottom-up" theories of the RT-IQ correlation. Different
researchers prefer one or the other, but the issue has not been definitively
decided empirically.
238 Intelligence

T. E. Reed and I had hoped that by finding a correlation between nerve

conduction velocity (NCV) in the brain's visual tract, from the retina to the
visual cortex, it would rule out the top-down theory, because the NCV is
measured long before the neural impulse has reached the higher association
centers that are necessarily involved in the kind of knowledge retrieval or
problem solving typically demanded by untimed psychometric tests. There was
indeed a significant correlation (-.27, corrected for range restriction in a
college sample, -.38) between individual differences in NCV and IQ (Reed &
Jensen, 1992). This finding clearly supports the "bottom-up" hypothesis, at
least as regards NCV and IQ. But alas, it does not enlighten the issue regarding
RT and IQ, because we found that the measure of NCV in the visual tract is
not correlated with RT as measured by the Hick or the odd-man-out
procedures (Reed & Jensen, 1993). This puzzle suggests new hypotheses, but
there is as yet no compelling explanation. Speculation should be postponed,
however, until a replication of these results insures their reliability.

3. CONCLUSIONS AND PERSPECTIVE

Clearly, much remains to be learned about the nature of g through further

investigations into the causes of its relation to RT. Using a variety of
experimentally manipulated ECTs to achieve the maximum possible correla-
tions between RT and g, it is then possible to investigate the physiological
correlates of these RTs with measures of NCV, AEP, and glucose metabolic
rate in localized regions of the brain. There is good reason to believe that a
program of research utilizing such techniques carried out by a number of
independent laboratories dedicated to this common goal will, in the foresee-
able future, realize what Spearman (1927) envisaged as the aim of research on
human intelligence: "The final word on the physiological side of the problem
[of g] ... must come from the most profound and detailed direct study of the
human brain in its purely physical and chemical aspects."
I conclude by noting that it was entirely through Eysenck's influence that I
began reading Gallon and Spearman in the first place, and it is exceedingly
improbable that my two decades of researching the connection between RT
and g would ever have ensued had I not once heard Eysenck talk about Roth's
experiment with the Hick paradigm and caught some of his enthusiasm for the
subject.

REFERENCES

Eysenck, H. J. (1967). Intelligence assessment: A theoretical and experimental

approach. British Journal of Educational Psychology, 37, 81-98.
 The psychometrics of intelligence 239

Eysenck, H. J. (1994). A biological theory of intelligence. In D. K. Detterman (Ed.),

Current topics in human intelligence, Vol. 4, Theories of intelligence (pp. 117-149).
Norwood, NJ: Ablex.
Frearson, W. M., & Eysenck, H. J. (1986). Intelligence, reaction time (RT) and a new
"odd-man-out" RT paradigm. Personality and Individual Differences, 7, 807-817.
Jensen, A. R. (1982). Reaction time and psychometric g. In H. J. Eysenck (Ed.), A
model for intelligence (pp. 93-132). Berlin: Springer.
Jensen, A. R. (1985). The nature of the black-white difference on various psychometric
tests: Spearman's hypothesis. Behavioral and Brain Sciences, 8, 193-219.
Jensen, A. R. (1987a). Individual differences in the Hick paradigm. In P. A. Vernon
(Ed.), Speed of information processing and intelligence (pp. 101-175). Norwood, NJ:
Ablex.
Jensen, A. R. (1987b). The g beyond factor analysis. In R. R. Ronning, J. A. Glover, J.
C. Conoley, & J. C. Witt (Eds.), The influence of cognitive psychology on testing (pp.
87-142). Hillsdale, NJ: Erlbaum.
Jensen, A. R. (1993a). Spearman's g: Links between psychometrics and biology. Annals
of the New York Academy of Sciences, 702, 103-129.
Jensen, A. R. (1993b). Spearman's hypothesis tested with chronometric information-
processing tasks. Intelligence, 17, 47-77.
Jensen, A. R. (1994). Phlogiston, animal magnetism, and intelligence. In D. K.
Detterman (Ed.), Current topics in human intelligence, Vol. 4, Theories of intelligence
(pp. 257-284). Norwood, NJ: Ablex.
Reed, T. E., & Jensen, A. R. (1992). Conduction velocity in a brain nerve pathway of
normal adults correlates with intelligence level. Intelligence, 16, 259-272.
Reed, T. E., & Jensen, A. R. (1993). Choice reaction time and visual pathway nerve
conduction velocity both correlate with intelligence but appear not to correlate with
each other: Implications for information processing. Intelligence, 17, 191-203.
Roth, E. (1964). Die Geschwindikeit der Verarbeitung von Information und ihr
Zusammenhang mit Intelligenz. Zeitschrift fur Experimentelle und Angewandte
Psychologie, 11, 616-622.
Spearman, C. (1927). The abilities of man. London: Macmillan.
Chapter 12
Behavioral genetic and biological approaches to
intelligence

P. A. Vernon

1. INTRODUCTION

Throughout his career, Hans Eysenck has unceasingly championed and promoted
the view that individual differences in human intelligence can only be fully
understood through investigations of their genetic and biological bases. In this
chapter, some of the research that has pursued these investigations is presented
and it is concluded that Eysenck's pioneering attempts to encourage exploration
in these areas were both well-founded and have been amply rewarded.

2. THE HERITABILITY OF INTELLIGENCE

Eysenck (1993) distinguishes between three types of intelligence: social or

practical intelligence—which is manifested in the ways in which people use
their abilities in everyday situations and activities—psychometric intelligence—
referring to the abilities measured by standard tests of intelligence or IQ—and
biological intelligence (see Figure 11.1 in chapter 11). Biological intelligence is
viewed by Eysenck as being largely determined by genetic factors but not to be
wholly innate, nutrition being just one of a number of nongenetic factors that
might influence its development.
Eysenck (1979) and Fulker and Eysenck (1979a, 1979b) provide an early
overview of behavior genetic studies of intelligence. In these chapters and
elsewhere, Eysenck recognizes that there is abundant evidence demonstrating
the importance of both genetic and nongenetic factors (nature and nurture) in
determining individual differences in intelligence. Fulker and Eysenck (1979a)
examine a number of twin, sibling, and adoption studies conducted up to that
time and conclude that the best estimate of the heritability of intelligence—
that is, the proportion of the total variance in intelligence that is attributable to
genetic variability—is .69. Thus, nongenetic, environmental factors are res-
ponsible for about a third of the total variance: .31.
 Behavioral genetic and biological approaches to intelligence 241

Fulker and Eysenck (1979a) also point out the distinction that behavioral
geneticists make between shared and nonshared environmental influences.
Shared environmental factors refer to those aspects of the environment that
family members (e.g., twins or siblings) have in common and that contribute to
their resembling, or correlating with, one another. Nonshared environmental
factors are those that influence one sibling but not the other: these factors will
serve to make siblings dissimilar and, indeed, are the major contributors to the
differences that exist between identical twins in their personalities and levels of
intelligence (Plomin, DeFries, & McClearn, 1990; Vernon, Jang, Harris, &
McCarthy, 1997). From their review of the literature, Fulker and Eysenck
estimate shared environmental influences to account for 18% of the variance in
intelligence and nonshared environmental factors to account for the remaining
13% (part of this 13%, however, merely reflecting the less than perfect
reliability of the measures of intelligence that were used).
It is instructive to compare Fulker and Eysenck's (1979a) heritability and
environmentality estimates with those of more recent behavioral genetic
studies. Casto, DeFries, and Fulker (1995), for example, present estimates of
the contributions of genetic and shared and nonshared environmental
influences to three factors extracted from the Wechsler Intelligence Scale for
Children-Revised (WISC-R; Wechsler, 1974): Verbal Comprehension,
Perceptual Organization, and Freedom from Distractibility. Their estimates,
denoted h2 (for genetic influences), c2 (for shared environmental influences),
and e2 (for nonshared environmental influences), were based on data collected
from 574 pairs of twins. For Verbal Comprehension, h2 = .44, c2 = .31, and e2
= .24. For Perceptual Organization, h2 = .50, c2 = .16, and el = .34. For
Freedom from Distractibility, hi = .49, c2 = .19, and el = .32.
These heritability values are lower than the .69 estimated by Fulker and
Eysenck (1979a), though theirs referred to the heritability of general or full-
scale intelligence rather than to subfactors. However, other more recent
estimates of the heritability of full-scale IQ scores also tend to favor the
somewhat lower values, yet still suggesting that genetic factors account for a
sizable 40-60% of the variance (e.g., Chipuer, Rovine, & Plomin, 1990; Plomin
et al., 1990). Moreover, as Bouchard (1993) points out, simply estimating the
heritability of intelligence or other mental ability measures should no longer be
the goal of a behavioral genetic investigation. There is no longer any question
that genetic factors make a major contribution to individual differences in
intelligence, and many other more complex and interesting questions, some of
which Eysenck (1979) anticipated, can now be addressed. These include the
identification of the precise types of gene action that are involved (for example,
additive, dominant, or epistatic); the influence of assortative mating; the
specific nature of environmental factors and the joint effects of genetic and
environmental influences; and the role of multivariate analyses that can tease
apart the relative influence of genetic and nongenetic factors on the
242 Intelligence

covariation of two or more variables (Bouchard, 1993; Casto et al., 1995). A

detailed coverage of these topics is beyond the scope of this chapter, but may
be found in such sources as Bouchard (1993) and Plomin et al. (1990).

3. INTELLIGENCE AND REACTION TIMES

Following a long period during which reaction times were widely regarded as
having been shown to be unrelated to intelligence or other higher cognitive
functions, Eysenck (1953, 1967) was one of the first to revive interest in their
use by arguing that individual differences in intelligence were to a large extent
dependent on mental speed. Since this time, numerous studies have
demonstrated that Eysenck was quite correct about this (e.g., see Vernon,
1987, and chapter 11). Reaction times have also, since the time of Gallon, been
viewed by many as a more-or-less direct measure of biological intelligence
(Eysenck, 1993) and, as such, deserve some attention in this chapter.
Many different tasks have been employed to measure reaction times (RTs),
ranging from quite simple tasks such as Jensen's Hick paradigm (e.g., Jensen,
1987, and chapter 11) to tasks involving substantially more complex
information processing. What most of these tasks have in common, however,
are a low error-rate—subjects typically respond correctly on the great majority
of trials, such that the main feature distinguishing between their performance
is speed rather than accuracy—and the means to provide a measure of the
speed with which individuals can perform quite specific cognitive processes: for
example, how quickly they can apprehend, encode or retrieve different kinds of
information in short- or long-term memory.
When administered individually, RT or speed-of-processing tests correlate
between approximately —.20 and —.40 with scores on standard measures of
intelligence (Vernon, 1990); higher IQ scores being associated with faster, or
shorter, RTs. Typically, more complex RT tests correlate more highly with IQs
than less complex RT tasks: Eysenck's "odd-man-out" test (Frearson, Barrett,
& Eysenck, 1988; Frearson & Eysenck, 1986), for example, yields RTs which
correlate between —.48 to -.62 with IQ scores. A pronounced relationship (r =
.97) also exists between the relative complexity of different RT tests and the
degree to which these discriminate between groups of different mean IQ levels
(Vernon, Nador, & Kantor, 1985).
Studies that have administered multiple RT measures to subjects have
reported sizable multiple correlations between these and IQ scores. Vernon
(1990), for example, summarizes six studies in which IQs were regressed on
batteries of five or more RT tests, yielding shrunken multiple correlations
between .37 and .74. This is strong support for the role that Eysenck ascribes to
mental speed: the speed with which individuals can process different kinds of
information accounting for as much as 50% of the variance in intelligence.
 Behavioral genetic and biological approaches to intelligence 243

Jensen (1982) and Vernon (1985) have proposed that speed-of-processing is

a property of the short-term or working-memory system that allows it to cope
with the information-processing demands of complex tasks (such as IQ tests)
despite the fact that the system has a limited storage capacity and is unable to
store information for more than a short period of time in the absence of
continual rehearsal. Eysenck (1993) favors a more biological explanation of the
speed—IQ relationship that focuses on the probability of error of transmission
of impulses across synapses: more error-free processing allowing both faster
information-processing speed and higher intelligence.
As Eysenck (1993) expected, there is evidence relating individual differences
in speed-of-processing to underlying biological mechanisms. McGue,
Bouchard, Lykken, and Feuer (1984), for example, reported that an overall
speed of response factor had a heritability of .46. Similar results were found by
Ho, Baker, and Decker (1988) and Vernon (1989), who reported heritabilities
between .49 and .52 for different RT measures.
Closer examination of the heritabilities of the different speed-of-processing
measures or parameters reported by McGue et al. (1984) and by Vernon
(1989) reveals quite pronounced variability in their magnitudes: ranging from
.116 to .604 across nine parameters derived from three tasks in McGue et al.
(1984) and from .24 to .90 across 11 RT measures in Vernon (1989). There are
also differences in the degree to which these different RT measures correlate
with IQ and importantly, in both studies, there is a positive relationship
between the RT measures' heritabilities and their correlations with IQ: r = .48
in McGue et al. (1984) and r = .61 in Vernon (1989). Thus, the more heritable
a RT measure, the stronger its correlation with IQ, or, from the other
perspective, the moreg-loaded a test—even if it is a relatively simple RT test—
the more highly heritable that test is.
Behavior genetics methodology allows one not only to compute heritabilities
of variables but also to estimate the extent to which phenotypic correlations
between two or more variables represent correlated genetic factors and/or
correlated environmental factors (Neale & Cardon, 1992). That is, is it the case
that IQ scores and RTs are correlated because both are dependent on the
same genes—do the same biological or physiological factors operate on both
intelligence and on speed of information-processing—and/or are there certain
environmental factors that influence both an individual's IQ and the speed
with which he/she can process different kinds of information? Multivariate
analyses of twin or other genetically-informative data allow these questions to
be answered.
In Ho et al. (1988), phenotypic correlations between full-scale IQ scores and
each of two speed-of-processing measures that they employed were approxi-
mately .40. For one of the speed measures—a rapid automatic naming task—
multivariate biometrical analyses revealed that 70% of this phenotypic
correlation with IQ could be explained by correlated genetic effects and the
244 Intelligence

remaining 30% by correlated specific environmental effects. The phenotypic

correlation between a second speed measure—the Colorado perceptual speed
task—and IQ was entirely attributable to correlated genetic effects. Similarly,
in a reanalysis of Vernon's (1989) twin data, Baker, Vernon, and Ho (1991)
reported phenotypic and genetic correlations between Verbal and
Performance IQ scores and general speed-of-processing factor scores derived
from eight RT measures. In this study, phenotypic speed-VIQ and speed-PIQ
correlations were both approximately .60 and a very sizable proportion of these
correlations was found to be attributable to correlated genetic factors: genetic
correlations between verbal IQ and RTs being 1.00 and between performance
IQs and RTs being .92. Finally, Rijsdijk, Vernon, and Boomsma (in press)
reported that phenotypic correlations between RTs and measures of intelli-
gence obtained from some 200 pairs of twins at age 16 and again at age 17 1/2
were entirely determined by common genetic influences.
These results indicate that genetic variations which lead to faster
information processing are highly associated with genetic variations important
to high IQ scores. They further suggest that mental speed and IQ may share
common biological mechanisms. Research that has sought to identify
biological factors that correlate with either intelligence or with speed of
information processing or with both is discussed in the next section.

4. BIOLOGICAL CORRELATES OF INTELLIGENCE

Over the course of this century, various researchers have attempted to identify
the biological basis of human intelligence. Some of these attempts have seemed
fruitful at first but have either failed to replicate or have not been followed up
for one reason or another; others have yielded a more consistent pattern of
results. In this section, three approaches to the investigation of biological
correlates of intelligence are described briefly—averaged evoked potentials,
cerebral glucose metabolic rates, and nerve conduction velocity. A fuller
account of these and other biological mechanisms is given in Vernon (1993b).

4.1 Averaged evoked potentials

In the 1960s, a number of investigators looked at relationships between
intelligence and spontaneous EEGs (e.g., Ellingson, 1966; Giannitrapani, 1969;
Vogel & Broverman, 1964, 1966). The results of these studies were somewhat
inconclusive (Gale & Edwards, 1983; but see also Anokhin & Vogel, 1996),
prompting other researchers such as Ertl and his co-workers to study various
parameters of evoked potentials (e.g., Chalke & Ertl, 1965; Ertl & Schafer,
1969).
 Behavioral genetic and biological approaches to intelligence 245

As their name implies, evoked potentials refer to the electrical activity of the
brain that is evoked by some external stimulus (e.g., a light flash or an auditory
stimulus such as a beep or a click). Any individual potential may show
considerable random fluctuations but over a large number of trials these
fluctuations can be smoothed out, yielding an averaged evoked potential
(AEP). Two parameters of AEPs have been studied extensively: their laten-
cies—measuring the speed with which the brain responds to the stimulus—and
their amplitude—measuring the amount of electrocortical activity that the
stimulus evokes.
Ertl and Schafer (1969) provided an early demonstration of correlations
between AEP parameters and measures of intelligence. In this study, several
IQ tests—including the Wechsler Intelligence Scale for Children, the Primary
Mental Abilities Test, and the Otis—were administered to 573 primary school
children. Correlations between IQs and AEP latencies were primarily negative
(ranging from .10 to —.35), indicating that higher IQ scores were associated
with shorter, or faster, latencies.
Following Ertl, a number of other AEP-IQ studies were conducted, many of
which reported significant correlations between IQ and AEP latencies and/or
amplitudes (e.g., Gucker, 1973; Plum, 1968; Rhodes, Dustman, & Beck, 1969;
Shucard & Horn; 1972, 1973; Weinberg, 1969). Most of the correlations in
these studies were of about the same magnitude as those obtained by Ertl and
Schafer (1969). More recent studies have provided more fine-grained estimates
of different AEP components, many of which show significant and sometimes
quite pronounced correlations with measures of both intelligence and speed of
information processing (e.g., Caryl, 1994; McGarry-Roberts, Stelmack, &
Campbell, 1992; Neubauer, Freudenthaler, & Pfurtscheller, 1995).
Other recent studies have focused on what has become known as the "string
measure" (A. E. Hendrickson, 1982; D. E. Hendrickson, 1982; D. E.
Hendrickson & A. E. Hendrickson, 1980). This measure—essentially, the
length of the contour perimeter of the AEP waveform—has yielded variable
results, some studies reporting substantial correlations between it and intelli-
gence (e.g., Blinkhorn & Hendrickson, 1982; Caryl & Fraser, 1985; Eysenck &
Barrett, 1984; Haier, Robinson, Braden, & Williams, 1983; Hendrickson, 1982;
Stough, Nettelbeck, & Cooper, 1990), others reporting nonsignificant
correlations or even correlations in the opposite direction to prediction (e.g.,
Barrett & Eysenck, 1992; Bates & Eysenck, 1993; Shagrass, Roemer,
Straumanis, & Josiassen, 1981). A recent article by Bates, Stough, Mangan,
and Pellet (1995) provides some resolution to the inconsistency of the results
and argues that positive and negative correlations between intelligence and
string length measures derived from different tasks can both be expected,
depending upon the attentional demands of the task.
246 Intelligence

The Hendrickson's have interpreted the high string length-IQ correlation in

terms of individual differences in the probability of occurrence of errors in the
transmission of pulses at the neuronal level. As mentioned earlier, Eysenck
(1982, 1993) has extended this error theory to account for the finding of
correlations between IQs and cognitive measures of speed of information
processing. As Deary and Caryl (1993) point out, however, the generally
positive empirical results obtained with the string length, and in studies
relating other AEP parameters to intelligence, have not yet led to the
development of a satisfactory theory. Most promising, perhaps, is Schafer's
(1979, 1982) model of neural adaptability, which, like the neural efficiency
model (Haier et al., 1988; Vernon, 1985, 1993a), proposes that the brains of
individuals of higher intelligence will respond to stimuli, process information,
and perform cognitive operations more quickly and efficiently than the brains
of persons lower in intelligence.

4.2 Cerebral glucose metabolism

The active brain, like any physical organ, consumes energy. Thus, if individuals
are engaged in any task that requires cognitive activity, an index of the extent to
which their brain is "working" is the rate at which it metabolizes glucose to
compensate for its expenditure of energy. The cerebral glucose metabolic rate
(CGMR) can be measured by means of positron emission tomography (PET)
scans, which map the concentration of a positron-emitting radionucleide that
has been intravenously administered to subjects and which is taken up into the
brain over a period of seconds or minutes, depending on the specific tracer.
The isotope concentrations measured by the PET scans are then converted
into estimates of CGMR.
A number of early PET scan studies involved Alzheimer patients or subjects
with other forms of dementia. Chase, Fedio, Brooks, Foster, Di Chiro, and
Mansi (1984), for example, obtained WAIS IQs from 17 Alzheimer patients
and five controls of normal intelligence and then measured their CGMRs
while they were at rest. These CGMRs correlated .61, .56, and .68 with Verbal,
Performance, and Full-Scale IQs, respectively. Other studies have also report-
ed a decline in the CGMRs of Alzheimer patients (e.g., Benson, 1982; Benson,
Metter, Kuhl, & Phelps, 1982; de Leon et al., 1983; Martin et al., 1986), and
Alavi et al. (1980) reported that the CGMRs of six senile dementia patients
were approximately 20-30% lower than those of four age-matched normal
controls, whose CGMRs were themselves some 50% lower than those of eight
young normal subjects.
When subjects are engaged in some cognitive task during the tracer uptake
period, a different pattern of results emerges. Haier et al. (1988), for example,
engaged 30 normal subjects in either a continuous performance task (CPT) or
the Advanced Raven Matrices and then obtained PET scans. Subjects who had
 Behavioral genetic and biological approaches to intelligence 247

worked on the Raven showed greater relative CGMRs than the CPT subjects;
thus, as expected, the heavier cognitive demands of the more challenging
Raven resulted in a greater expenditure of energy. Within the Raven subjects,
however, high negative correlations, between —.44 and —.84, were found
between Raven scores and absolute GMRS; indicating that subjects who
obtained the highest Raven scores actually consumed the least amount of
energy. These results should be interpreted with caution because of the small
sample size. The results have, however, been replicated: Parks et al. (1988)
reporting correlations of approximately -.50 between relative CGMRs and
verbal fluency test scores in a sample of 16 normal subjects, Haier et al. (1995)
reporting a correlation of —.58 between whole brain CGMRs and WAIS-R IQs
among samples of mildly mentally retarded, Down's syndrome, and normal
subjects; and Berent et al. (1988) and Boivin et al. (1992) also reporting
negative correlations between CGMRs and tests of cognitive function.
Taken together, the PET scan studies provide compelling evidence that, at
rest, when subjects can engage in any mental activity they wish, those subjects
with higher IQs demonstrate increased brain activity but, when they are
required to perform an assigned cognitive task, subjects with higher IQs are
able to accomplish the task with a lower consumption of energy. Subjects of
higher IQ can thus be characterized as having greater "brainpower" at their
disposal while at the same time being able to use their brainpower more
efficiently when called upon to do so (though see Larson, Haier, LaCasse, &
Hazen, 1995). The use of PET scans remains an expensive undertaking—Haier
(1990) states that an eight-subject PET scan study may cost as much as
$20,000—but this is offset by the tremendous potential that these studies have
to yield information about the biological basis of intelligence.

4.3 Nerve conduction velocity

Nerve conduction velocity (NCV) refers to the speed with which electrical
impulses are transmitted along nerve fibers and across synapses. Reed (1984)
proposed that individual differences in NCV might be attributable to genetic
differences in the structure and amount of transmission proteins which, in turn,
set limits on information-processing rates. Thus, faster NCV would be
expected to correlate with faster RTs and higher intelligence.
Since Reed's 1984 paper, a number of researchers have investigated
relationships between NCV, RTs, and IQ. Unfortunately, there has been no
agreement in the results. Veraon and Mori (1992), for example, reported
significant correlations between IQs, RTs, and peripheral NCVs (measured in
the median nerve of the arm) in two samples of university students. NCV-IQ
correlations were .42 and .48; NCV-RT correlations were —.28 and —.18.
Thus, as expected, faster NCVs were associated with higher IQs and faster
speed of information processing. Wickett and Vernon (1994), however, failed
248 Intelligence

to replicate these results, despite using the same procedure to measure NCV.
In this study, two measures of NCVs correlated —.12 and .02 with IQs and .02
and .01 with RTs.
Other investigators have also failed to find any correlation between IQs and
NCV. These include Barrett, Daum, and Eysenck (1990), Reed and Jensen
(1991), and Rijsdijk, Boomsma, and Vernon (1995). This latter study, however,
did report that peripheral NCV in humans is a substantially heritable trait (h2
= .77), and Rijsdijk and Boomsma (1997) not only found significant positive
peripheral NCV-IQ correlations, of the order of approximately .15, but also
reported that these phenotypic NCV-IQ correlations were entirely mediated
by common genetic factors. Moreover, although Reed and Jensen (1991) did
not find a significant correlation between peripheral NCV and IQ, Reed and
Jensen (1992) reported a significant correlation between IQ and NCV in a
brain nerve pathway.
Procedural differences between studies may account for some of the
discrepancies between their results. In addition, Wickett and Vernon's (1994)
reanalysis of Vernon and Mori's (1992) data suggested that higher IQ-NCV
correlations occur in males than in females. Vernon and Wickett (1997) are
currently completing a large-scale study of IQs, RTs, and peripheral and
central NCVs which will address some of the outstanding questions that
remain to be answered concerning relationships between these variables. Even
if it is granted that NCVs, RTs, and IQ are intercorrelated (and at this point
this cannot be considered established), the pattern of the relationship between
them does not appear to follow predictions. Vernon and Mori (1992), for
example, predicted that the partial correlation between IQs and RTs,
controlling for NCV, should be substantially smaller than their zero-order
correlation, but this was not supported. Anderson (1994) also suggests that
relationships between IQs and RTs cannot be attributed to speed of neuron
conduction. At present, the IQ-NCV studies that have been conducted have
probably raised more questions than they have answered, but this is neither
necessarily a bad thing, providing that the questions which are raised are
testable, nor unique to these particular variables.

5. INTELLIGENCE AND PERSONALITY

Eysenck's models of intelligence have long included noncognitive components

of personality, such as continuance (or persistence) and error-checking (or
impulsivity). Numerous behavioral genetic studies have demonstrated that
different dimensions of personality are heritable; most showing heritabilities of
about .50 (Plomin et al., 1990). A recent study by Harris, Vernon, and Jang (in
press) performed univariate and multivariate analyses on measures of
 Behavioral genetic and biological approaches to intelligence 249

personality and intelligence that had been completed by samples of adult twins:
92 MZ pairs and 50 same-sex DZ pairs.
The measure of personality was the Personality Research Form-E (PRF-E;
Jackson, 1989), an inventory which assesses 20 personality traits. Subjects were
also administered the Multidimensional Aptitude Battery (MAB; Jackson,
1984), a timed, group-administered intelligence test yielding full-scale, verbal,
and performance IQ scores.
At the phenotypic level, significant zero-order correlations indicated that
intelligence was positively correlated with the personality traits achievement,
autonomy, dominance, endurance, sentience, and understanding, and
negatively related to harm avoidance, order, and succorance. Univariate gene-
tic analyses revealed that, as expected, the intelligence and personality scales
were heritable: full-scale IQs having a heritability of .49 and the median
heritability of the 20 PRF-E scales being .51. Finally, multivariate genetic
analyses revealed moderate to quite sizable genetic correlations between
intelligence and personality, indicating that the phenotypic correlations
between the two were largely attributable to the influence of common genetic
factors. Referring back to Eysenck's model of intelligence, it is noteworthy that
the PRF-E endurance scale—which taps persistence and continuance—
showed significant phenotypic and genetic correlations with IQs. The PRF-E
also has an impulsivity scale: this scale showed no phenotypic correlation with
IQ in Harris et al. (in press) but did show a small genetic correlation (r = .16).

6. NEW RESEARCH DIRECTIONS IN BEHAVIOR GENETICS

Recent developments in behavioral genetic methodology have greatly extended

the range and types of research questions that can be answered with
genetically-informative data sets. In this section, two such developments will
be described to illustrate some of these advances: the concept of group
heritability and the molecular genetic identification of quantitative trait loci.
These topics are just two of many examples that could be chosen to illustrate
the new areas of inquiry that behavior geneticists are now exploring; additional
examples may be found in Neale and Cardon (1992).

6.1 Group heritability

While individual heritability, as described above, refers to the extent to which
individual differences on some trait are attributable to genetic differences
between people, the group heritability of a trait indicates the extent to which
the difference between the population mean and persons who score at one
extreme or the other on the trait can be accounted for by genetic differences.
Considering the mentally retarded as an example, an estimate of group
250 Intelligence

heritability would address the question of whether the same genetic and
environmental factors that contribute to individual differences in intelligence
are also responsible for their extreme IQ scores, or whether different genetic
(and environmental) factors contribute to the etiology of the difference
between their level of intelligence and the mean of the distribution.
The group heritability of a trait can be estimated by comparing MZ and DZ
twins, at least some of whom are discordant on the trait: one member of each
twin pair obtaining an extreme score on the trait and the other obtaining either
an equally extreme or a less extreme or normal score. A comparison is then
made between the extent to which the average scores of the MZ and DZ
cotwins of the extreme scorers regress back to the population mean of the trait.
If these MZ and DZ cotwins regress back to the mean by the same amount, it
may be inferred that the group heritability of the trait is zero: none of the
differences between the extreme scorers and the population mean can be
attributed to genetic factors. However, if the MZ cotwins obtain the same
mean as their extreme-scoring twins while the DZ cotwins regress halfway back
to the population mean, the group heritability of the trait would be 1.0.
Plomin (1991) presents an illustrative example of group heritability,
estimated from twin data collected in a study of reading disabilities by
DeFries and Fulker (1985). In this study, the mean combined reading,
perceptual speed, and memory scores of MZ and DZ twin-pairs, at least one of
whom had a diagnosed reading disability, were compared to the mean of
nontwin controls. The reading disabled twins represent the extreme scorers in
this study, their cotwins may or not also have been reading disabled, and the
nontwin controls represent the normal population.
Setting the population (control) mean at 0, and expressing the scores of the
twins as standardized deviations from this mean, the diagnosed reading-
disabled MZ and DZ twins obtained a mean of -2.8, while the means of their
MZ and DZ cotwins were —2.0 and —1.6, respectively. Noting that the DZ
cotwin mean has regressed closer to the population mean than the MZ cotwin
mean, Plomin first computes the ratios 2.0/2.8 = .71 for the MZ twins, and 1.6/
2.8 = .57 for the DZ twins. Doubling the difference between these MZ and DZ
ratios provides the estimate of group heritability. Thus, the group heritability is
2(.71-.57) = .28, indicating that 28% of the difference between the scores of
the diagnosed reading disabled twins and the general population is attributable
to genetic factors. Common environmental factors account for .71-.28 = 43%
of the difference, and unique or nonshared environmental factors account for
the remaining 29% of the difference. Finally, the fact that the group heritability
of .28 is different from the individual heritability of reading scores, estimated
by Plomin to be .50, indicates that the genetic etiology of reading disabilities is
different from that of individual differences in the normal distribution of
reading ability.
 Behavioral genetic and biological approaches to intelligence 251

Group heritability estimates thus reveal both the extent to which differences
between extreme scorers and the rest of the distribution are attributable to
genetic factors, and the extent to which these genetic factors are the same or
different from those that contribute to individual differences. This information,
in turn, may serve as a useful starting point in the investigation and
identification of the specific genes (and the specific environmental factors)
that are involved, one approach to which is addressed in the following section.

6.2 Quantitative trait loci

One of the most exciting new developments in behavioral genetic research has
been the beginning of the search for specific genes that contribute to variability
on different traits, including intelligence (Plomin et al., 1994, 1995; Skuder et
al., 1995). Starting as recently as 1994, Plomin and his co-workers have
presented preliminary results of the IQ QTL Project, where QTL refers to the
quantitative trait loci or the multiple genes that are presumed to have some
effect, of greater or lesser size, on intelligence.
Because QTL which exert only a small effect are unlikely to be identified by
traditional linkage strategies (Plomin, 1990), the IQ QTL Project has instead
adopted an allelic association strategy. The frequencies of alleles in groups of
very high and very low IQ subjects are compared in an attempt to identify one
or more genes that occur more frequently in one group than the other. As an
example of this approach, Plomin et al. (1994) refer to the recent identification
of a gene associated with late-onset Alzheimer's disease: Corder et al. (1993)
reported that the E4 allele of apolipoprotein E occurs in 64% of sporadic
Alzheimer's cases, in 80% of familial Alzheimer's cases, and in only 31% of
control groups.
In Plomin et al. (1994), an original and a replication sample were selected.
The original sample consisted of 24 high-IQ, 21 middle-IQ, and 18 low-IQ
subjects. The mean IQs of the high- and low-IQ groups (130 and 82,
respectively) put them in about the top or bottom 5% of the population. The
replication sample had two even more extreme groups: 27 high-IQ subjects
(mean IQ = 142) and 17 low-IQ subjects (mean IQ = 59), representing about
the top or bottom 1% of the normal distribution. All low-IQ subjects were
screened for perinatal problems, accidents, or illnesses that might have
contributed to their low intelligence.
The high- and low-IQ groups were genotyped for 46 two-allele DNA
markers and for 14 multiple-allele DNA markers, all of which were in or near
to genes thought to be relevant to neural functioning. Comparisons in the
original sample revealed that five of the 46 two-allele markers occurred with
significantly (p < .05) different frequencies in the high- and low-IQ groups.
Three of these comparisons failed to replicate in the replication sample but two
others showed differences which, although not statistically significant, were in
252 Intelligence

the same direction as in the original sample. Combining the two high and the
two low-IQ groups from the original and replication samples yielded significant
differences between the frequencies of these two markers: HLA-A(B) (p =
.006) and SOD2(B) (p = .03). Turning to the multiple-allele markers, three of
26 comparisons between the high- and low-IQ groups in the original sample
were significant (p < .05), although none of these were replicated in the
replication sample. However, another allele—CTGB33—showed a marginally
significant effect (p = .052) in the original sample and a significant difference
(p = .048) for this same allele was also observed in the replication sample.
Combining the samples showed allelic frequencies for CTGB33 of .34 in the
high-IQ group and .56 in the low-IQ group: a highly significant difference (p =
.006). Moreover, this marker is of particular interest because it is expressed in
the brain.
Plomin et al. (1995) studied 40 additional DNA markers in the same high-
and low-IQ groups. In this investigation, three new markers showed significant
(p < .05) allelic frequency differences between the original high- and low-IQ
groups. One of these (EST00083) replicated significantly in the replication
sample, and the other two (ADH5 and NGFB) showed frequencies in the same
direction but did not reach statistical significance. A more detailed discussion
of the significant finding with EST00083 is provided in Skuder et al. (1995).
Briefly, this marker involves mitochondrial rather than nuclear DNA and its
unexpected nature led the researchers to express caution in identifying it as a
QTL for IQ. This caution may have been well-grounded, another research
team having failed to replicate the result in an independent sample, albeit
using markedly different measures of intelligence (Moises et al., in press).
It is to be hoped that additional research teams will continue to undertake
QTL projects. Of course, these need not be restricted only to comparisons
between groups of high and low intelligence, but the early results from the IQ
QTL project are sufficiently promising to encourage further investigation of
this sort. QTL research into the genetic basis of variability in intelligence has
not gone without criticism (e.g., Weiss, 1995), and others have expressed
concern about ethical considerations that may need to be taken into account
(Harper, 1995). These considerations notwithstanding, QTL studies represent
one of the most promising approaches to the molecular genetic investigation of
human intelligence and it seems likely that they will be of considerable benefit
in advancing our understanding of the genetic basis of individual differences in
this and other complex behavioral traits.

7. CONCLUSIONS

Hans Eysenck has made, and continues to make, many important empirical
and theoretical contributions to the area of human intelligence. His positions
 Behavioral genetic and biological approaches to intelligence 253

have sometimes been unpopular and hotly contested but overall the balance of
evidence has turned out to weigh in his favor more often than not. This is
perhaps particularly true in the areas of behavior genetics and biology of
intelligence. There are few (if any) serious scholars who still maintain that
individual differences in intelligence are not attributable to a sizable degree to
genetic factors; to assert otherwise requires a highly creative interpretation of
the evidence. There is also no longer any doubt, despite the once widespread
belief to the contrary, that the speed with which individuals can execute
different basic cognitive processes is moderately to quite highly correlated with
performance on much more complex tests of reasoning and intelligence. Speed
of information processing has also been shown to have a substantial heritable
component and there are an increasing number of laboratories finding
significant relationships between intelligence, cognitive abilities, and a number
of underlying biological mechanisms. New methodologies, such as QTL, have
the potential to advance our knowledge of the genetic basis of intelligence still
further.
Directly or indirectly, Eysenck has been a pioneer at the forefront of much
of the work described in this and the other chapters. This book is a fitting
testament to the services he has rendered his profession.

REFERENCES

Alavi, A., Ferris, S., Wolf, A., Reivich, M., Farkas, T., Dann, R., Christman, D.,
MacGregor, R. R., & Fowler, J. (1980). Determination of cerebral metabolism in
senile dementia using F-18 deoxyglucose metabolism and positron emission
tomography. Journal of Nuclear Medicine, 21, 21.
Anderson, B. (1994). Speed of neuron conduction is not the basis of the IQ-RT
correlation: Results from a simple neural model. Intelligence, 19, 317-323.
Anokhin, A., & Vogel, F. (1996). EEG alpha rhythm frequency and intelligence in
normal adults. Intelligence, 23, 1-14.
Baker, L. A., Vernon, P. A., & Ho, H. Z. (1991). The genetic correlation between
intelligence and speed of information-processing. Behavior Genetics, 21, 351-367.
Barrett, P. T., Daum, L, & Eysenck, H. J. (1990). Sensory nerve conduction and
intelligence: A methodological study. Journal of Psychophysiology, 4, 1-13.
Barrett, P., & Eysenck, H. J. (1992). Brain evoked potentials and intelligence: The
Hendrickson paradigm. Intelligence, 16, 361-382.
Bates, T., & Eysenck, H. J. (1993). String length, attention and intelligence: Focused
attention reverses the string length-IQ relationship. Personality and Individual
Differences, 15, 363-371.
Bates, T., Stough, C., Mangan, G., & Pellett, O. (1995). Intelligence and complexity of
the averaged evoked potential: An attentional theory. Intelligence, 20, 27-39.
Benson, D. F. (1982). The use of positron emission scanning techniques in the diagnosis
of Alzheimer's disease. In S. Corkin, K. L. Davis, J. H. Growdon, E. Vodin, & R. J.
254 Intelligence

Wurtman (Eds.), Alzheimer's disease: A review of progress in research. New York:

Raven Press.
Benson, D. F., Metter, E. J., Kuhl, D. E., & Phelps, M. E. (1982). Positron computed
tomography in neurobehavioral problems. In A. Kertesz (Ed.), Localization in
neuropsychology. New York: Academic Press.
Berent, S., Giordani, B., Lehtinen, S., Markel, D., Penney, J., Buchtel, H., Starosta-
Rubinstein, S., Hichwa, R., & Young, A. (1988). Positron emission tomographic scan
investigations of Huntington's disease: Cerebral metabolic correlates of cognitive
function. Annals of Neurology, 23, 541-546.
Blinkhorn, S. F., & Hendrickson, D. E. (1982). Averaged evoked responses and
psychometric intelligence. Nature, 295, 596-597.
Boivin, M. J., Giordani, B., Berent, S., Amato, D. A., Lehtinen, S., Koeppe, R. A.,
Buchtel, H. Foster, N., & Kuhl, D. (1992). Verbal fluency and positron emission
tomographic mapping of regional cerebral glucose metabolism. Cortex, 28, 231-239.
Bouchard, T. J. (1993). The genetic architecture of human intelligence. In P. A. Vernon
(Ed.), Biological approaches to the study of human intelligence. Norwood, NJ: Ablex.
Caryl, P. G. (1994). Early event-related potentials correlate with inspection time and
intelligence. Intelligence, 18, 15-46.
Caryl, P. G., & Fraser, I. C. (1985). The Hendrickson string length and intelligence—a
replication. Paper presented at the Psychophysiology Society Scottish Conference,
Edinburgh.
Casto, S. D., DeFries, J. C., & Fulker, D. W. (1995). Multivariate genetic analysis of
Wechsler Intelligence Scale for Children-Revised (WISC-R) factors. Behavior
Genetics, 25, 25-32.
Chalke, F., & Ertl, J. (1965). Evoked potentials and intelligence. Life Science, 4, 1319-
1322.
Chase, T. N., Fedio, P., Brooks, R., Foster, N. L., Di Chiro, G., & Mansi, L. (1984).
Wechsler Adult Intelligence Scale performance: Cortical localization by fluorodeox-
yglucose F18-positron emission tomography. Archives of Neurology, 41, 1244-1247.
Chipuer, H. M., Rovine, M., & Plomin, R. (1990). LISREL modelling: Genetic and
environmental influences on IQ revisited. Intelligence, 14, 11-29.
Corder, E. H., Saunders, A. M., Strittmatter, W. J., Schmechel, D. E., Gaskell, P. C.,
Small, G. W., Roses, A. D., Haines, J. L., & Pericak-Vance, M. A. (1993). Gene dose
of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset
families. Science, 261, 921-923,
Deary, I. J., & Caryl, P. G. (1993). Intelligence, EEG, and evoked potentials. In P. A.
Vernon (Ed.), Biological approaches to the study of human intelligence. Norwood, NJ:
Ablex.
DeFries, J. C., & Fulker, D. W. (1985). Multiple regression analysis of twin data.
Behavior Genetics, 15, 467-473.
de Leon, M. J., Ferris, S. H., George, A. E., Christman, D. R., Fowler, J. S., Gentes, C.,
Reisberg, B., Gee, B., Emmerich, M., Yonekura, Y., Brodie, J., Kricheff, I. I., &
Wolf, A. P. (1983). Positron emission tomographic studies of aging and Alzheimer
disease. American Journal of Neuroradiology, 4, 568-571.
EUingson, R. J. (1966). Relationship between EEG and test intelligence: A
commentary. Psychological Bulletin, 65, 91-98.
 Behavioral genetic and biological approaches to intelligence 255

Ertl, J. P., & Schafer, E. W. P. (1969). Brain response correlates of psychometric

intelligence. Nature, 223, 421-422.
Eysenck, H. J. (1953). Uses and abuses of psychology. London: Pelican.
Eysenck, H. J. (1967). Intelligence assessment: a theoretical and experimental
approach. British Journal of Educational Psychology, 37, 81-98.
Eysenck, H. J. (1979). The structure and measurement of intelligence. Berlin: Springer.
Eysenck, H. J. (Ed.) (1982). A model for intelligence. Berlin: Springer.
Eysenck, H. J. (1993). The biological basis of intelligence. In P. A. Vernon (Ed.),
Biological approaches to the study of human intelligence. Norwood, NJ: Ablex.
Eysenck, H. J., & Barrett, P. (1984). Psychophysiology and the measurement of
intelligence. In C. R. Reynolds & V. Willson (Eds.), Methodological and statistical
advances in the study of individual differences. New York: Plenum Press.
Frearson, W., Barrett, P., & Eysenck, H. J. (1988). Intelligence, reaction tune, and the
effects of smoking. Personality and Individual Differences, 9, 497-517.
Frearson, W. M., & Eysenck, H. J. (1986). Intelligence, reaction time (RT) and a new
"odd-man-out" RT paradigm. Personality and Individual Differences, 7, 807-817.
Fulker, D. W., & Eysenck, H. J. (1979a). Nature and nurture: heredity. In H. J. Eysenck
(Ed.), The structure and measurement of intelligence. Berlin: Springer.
Fulker, D. W., & Eysenck, H. J. (1979b). Nature and nurture: environment. In H. J.
Eysenck (Ed.), The structure and measurement of intelligence. Berlin: Springer.
Gale, A., & Edwards, J. A. (1983). Cortical correlates of intelligence. In A. Gale & J. A.
Edwards (Eds.), Physiological correlates of human behaviour. Vol. 3: Individual
differences and psychopathology. London: Academic Press.
Giannitrapani, D. (1966). EEG average frequency and intelligence. Electroencephalo-
graphy and Clinical Neurophysiology, 27, 480-486.
Gucker, D. K. (1973). Correlating visual evoked potentials with psychometric
intelligence: Variation in technique. Perceptual and Motor Skills, 37, 189-190.
Haier, R. J. (1990). The end of intelligence research. Intelligence, 14, 371-374.
Haier, R. J., Chueh, D., Touchette, P., Lott, I., Buchsbaum, M. S., MacMillan, D.,
Sandman, C., LaCasse, L., & Sosa, E. (1995). Brain size and cerebral glucose
metabolic rate in nonspecific mental retardation and Down's syndrome. Intelligence,
20, 191-210.
Haier, R. J., Seigel, B. V., Nuechterlein, K. H., Hazlett, E., Wu, J. C., Paek, J.,
Browning, H. L., & Buchsbaum, M. S. (1988). Cortical glucose metabolic rate
correlates of abstract reasoning and attention studied with Positron Emission
Tomography. Intelligence, 12, 199-217.
Haier, R. J., Robinson, D. L., Braden, W., & Williams, D. (1983). Electrical potentials
of the cerebral cortex and psychometric intelligence. Personality and Individual
Differences, 4, 591-599.
Harper, P. S. (1995). DNA markers associated with high versus low IQ: Ethical
considerations. Letter to the editor of Behavior Genetics, 25, 197-198.
Harris, J. A., Vernon, P. A., & Jang, K. L. (in press). A multivariate genetic analysis of
relationships between intelligence and personality. Developmental Neuropsychology.
Hendrickson, A. E. (1982). The biological basis of intelligence. Part I: Theory. In H. J.
Eysenck (Ed.), A model for intelligence. Berlin: Springer.
Hendrickson, D. E. (1982). The biological basis of intelligence. Part II: Measurement.
In H. J. Eysenck (Ed.), A model for intelligence. Berlin: Springer.
256 Intelligence

Hendrickson, D. E., & Hendrickson, A. E. (1980). The biological basis of individual

differences in intelligence. Personality and Individual Differences, 1, 3-33.
Ho, H. Z., Baker, L., & Decker, S. N. (1988). Covariation between intelligence and
speed-of-processing: genetic and environmental influences. Behavior Genetics, 18,
247-261.
Jackson, D. N. (1984). Multidimensional aptitude battery manual. Port Huron, MI:
Research Psychologists Press.
Jackson, D. N. (1989). Personality research form manual (3rd ed.). Port Huron, MI:
Sigma Assessment Systems.
Jensen, A. R. (1982). Reaction time and psychometric g. In H. J. Eysenck (Ed.), A
model for intelligence. Berlin: Springer.
Jensen, A. R. (1987). Individual differences in the Hick paradigm. In P. A. Vernon
(Ed.), Biological approaches to the study of human intelligence. Norwood, NJ: Ablex.
Larson, G. E., Haier, R. J., LaCasse, L., & Hazen, K. (1995). Evaluation of a "mental
effort" hypothesis for correlations between cortical metabolism and intelligence.
Intelligence, 21, 267-278.
Martin, A., Brouwers, P., Lalonde, F., Cox, G, Teleska, P., Fedio, P., Foster, N. L., &
Chase, T. N. (1986). Towards a behavioral typology of Alzheimer patients. Journal of
Clinical and Experimental Neuropsychology, 8, 594-610.
McGarry-Roberts, P. A., Stelmack, R. M., & Campbell, K. B. (1992). Intelligence,
reaction time, and event-related potentials. Intelligence, 16, 289-313.
McGue, M., Bouchard, T. J., Lykken, D. T., & Feuer, D. (1984). Information processing
abilities in twins reared apart. Intelligence, 8, 239-258.
Moises, H. W., Yan, L., Neppert, M., Kohnke, P., Skuder, P., Petrill, S., & Plomin, R.
(in press). A mitochondrial DNA marker EST00083 is not associated with high
versus average IQ hi a German sample. Intelligence.
Neale, M. G, & Cardon, L. R. (1992). Methodology for genetic studies of twins and
families. Dordrecht: Kluwer Academic Publishers.
Neubauer, A. G, Freudenthaler, H. H., & Pfurtscheller, G. (1995). Intelligence and
spatiotemporal patterns of event-related desynchronization (ERD). Intelligence, 20,
249-266.
Parks, R. W., Loewenstein, D. A., Dodrill, K. L., Barker, W. W., Yoshii, F., Chang, J.
Y., Emran, A., Apicella, A., Sheramata, W. A., & Duara, R. (1988). Cerebral
metabolic effects of a verbal fluency test: A PET scan study. Journal of Clinical and
Experimental Neuropsychology, 10, 565-575,
Plomin, R. (1990). The role of inheritance in behavior. Science, 248, 183-188.
Plomin, R. (1991). Genetic risk factors and psychosocial disorders: Links between
normal and abnormal. In M. Rutter & P. Casaer (Eds.), Biological risk factors for
psychosocial disorders. Cambridge: Cambridge University Press.
Plomin, R., DeFries, J. G, & McClearn, G. E. (1990). Behavioral genetics: a primer (2nd
ed.). New York: Freeman.
Plomin, R., McClearn, G. E., Smith, D. L., Skuder, P., Vignetti, S., Choraey, M. J.,
Chorney, K., Kasarda, S., Thompson, L. A., Detterman, D. K., Petrill, S. A., Daniels,
J., Owen, M. J., & McGuffin, P. (1995). Allelic associations between 100 DNA
markers and high versus low IQ. Intelligence, 21, 31-48.
Plomin, R., McClearn, G. E., Smith, D. L., Vignetti, S., Chorney, M. J., Chorney, K.,
Venditti, C. P., Kasarda, S., Thompson, L. A., Detterman, D. K., Daniels, J., Owen,
 Behavioral genetic and biological approaches to intelligence 257

M., & McGuffin, P. (1994). DNA markers associated with high versus low IQ: The
IQ quantitative trait loci (QTL) project. Behavior Genetics, 24, 107-118.
Plum, A. (1968). Visual evoked responses: Their relationship to intelligence. Unpublished
doctoral dissertation, University of Florida, Gainesville.
Reed, T. E. (1984). Mechanism for heritability of intelligence. Nature, 311, 417.
Reed, T. E., & Jensen, A. R. (1991). Arm nerve conduction velocity (NCV), brain NCV,
reaction time, and intelligence. Intelligence, 15, 33-47.
Reed, T. E., & Jensen, A. R. (1992). Conduction velocity in a brain nerve pathway of
normal adults correlates with intelligence level. Intelligence, 16, 259-272.
Rhodes, L. E., Dustman, R. E., & Beck, E. C. (1969). The visual evoked response: A
comparison of bright and dull children. Electroencephalography and Clinical
Neurophysiology, 27, 364-372.
Rijsdijk, F. V., & Boomsma, D. I. (1997). Genetic mediation of the correlation between
peripheral nerve conduction velocity and IQ. Behavior Genetics, 27, 87-98.
Rijsdijk, F. V., Boomsma, D. I., & Vernon, P. A. (1995). Genetic analysis of peripheral
nerve conduction velocity in twins. Behavior Genetics, 25, 341-348.
Rijsdijk, F. V., Vernon, P. A., & Boomsma, D. I. (in press). The genetic basis of the
relation between speed of information-processing and IQ. Behavioral Brain Research.
Schafer, E. W. P. (1979). Cognitive neural adaptability: A biological basis for individual
differences in intelligence. Psychophysiology, 16, 199.
Schafer, E. W. P. (1982). Neural adaptability: A biological determinant of behavioral
intelligence. International Journal of Neuroscience, 17, 183-191.
Shagrass, C., Roemer, R. A., Straumanis, J. J., & Josiassen, R. C. (1981). Intelligence as
a factor in evoked potential studies of psychopathology 1. Comparison of low and
high IQ subjects. Biological Psychiatry, 16, 1007-1030.
Shucard, D. W., & Horn, J. L. (1972). Evoked cortical potentials and measurement of
human abilities. Journal of Comparative and Physiological Psychology, 78, 59-68.
Shucard, D. W., & Horn, J. L. (1973). Evoked potential amplitude change related to
intelligence and arousal. Psychophysiology, 10, 445-452.
Skuder, P., Plomin, R., McClearn, G. E., Smith, D. L., Vignetti, S., Chorney, M. J.,
Chorney, K., Kasarda, S., Thompson, L. A., Detterman, D. K., Petrill, S. A., Daniels,
J., Owen, M. J., & McGuffin, P. (1995). A polymorphism in mitochondrial DNA
associated with IQ? Intelligence, 21, 1-11.
Stough, C. K. K., Nettelbeck, T., & Cooper, C. J. (1990). Evoked brain potentials, string
length and intelligence. Personality and Individual Differences, 11, 401-406.
Vernon, P. A. (1985). Individual differences in general cognitive ability. In L. C.
Hartlage & C. F. Telzrow (Eds.), The neuropsychology of individual differences. New
York: Plenum.
Vernon, P. A. (Ed.) (1987). Speed of information-processing and intelligence. Norwood,
NJ: Ablex.
Vernon, P. A. (1989). The heritability of measures of speed of information-processing.
Personality and Individual Differences, 10, 573-576.
Vernon, P. A. (1990). An overview of chronometric measures of intelligence. School
Psychology Review, 19, 399-410.
Vernon, P. A. (1993a). Intelligence and neural efficiency. In D. K. Detterman (Ed),
Current topics in human intelligence (Vol. 3). Norwood, NJ: Ablex.
258 Intelligence

Vernon, P. A. (Ed.) (1993b). Biological approaches to the study of human intelligence.

Norwood, NJ: Ablex.
Vernon, P. A., Jang, K. L., Harris, J. A., & McCarthy, J. M. (1997). Environmental
predictors of personality differences: a twin and sibling study. Journal of Personality
and Social Psychology, 72, 177-183.
Vernon, P. A., & Mori, M. (1992). Intelligence, reaction times, and peripheral nerve
conduction velocity. Intelligence, 16, 273-288.
Vernon, P. A., Nador, S., & Kantor, L. (1985). Group differences in intelligence and
speed of information-processing. Intelligence, 9, 137-148.
Vernon, P. A., & Wickett, J. C. (1997). Nerve conduction velocity and MRI studies ofg.
Paper presented at the VIII International Conference of the International Society for
the Study of Individual Differences, Aarhus, Denmark, July.
Vogel, W., & Broverman, D. M. (1964). Relationship between EEG and test
intelligence: A critical review. Psychological Bulletin, 62, 132-144.
Vogel, W., & Broverman, D. M. (1966). A reply to "Relationship between EEG and test
intelligence: A commentary. Psychological Bulletin, 65, 99-109.
Wechsler, D. (1974). Examiner's Manual: Wechsler Intelligence Scale for Children (Rev.
ed.). San Antonio, TX: Psychological Corporation.
Weinberg, H. (1969). Correlation of frequency spectra of averaged visual evoked
potentials and verbal intelligence. Nature, 224, 813-815.
Weiss, V. (1995). The advent of a molecular genetics of general intelligence. Intelligence,
20, 115-124.
Wickett, J. C., & Vernon, P. A. (1994). Peripheral nerve conduction velocity, reaction
time, and intelligence: An attempt to replicate Vernon and Mori (1992). Intelligence,
18, 127-131.
Chapter 13
Geographical variation in intelligence

R. Lynn
1. INTRODUCTION

Geographical variation in intelligence is concerned with the way in which the

intelligence levels of different peoples vary systematically over the world with
geographical location, and with attempts to explain this variation. This chapter
considers these questions in eight sections. Section 1 consists of an outline of
the contributions of H. J. Eysenck to this issue. Section 2 presents a summary
of my work on the geography of intelligence up to 1990. Sections 3-5 consist of
summaries and discussions of recent studies published between 1990 and 1996
on the intelligence of the Mongoloid (principally Japanese and Chinese)
peoples in East Asia and in the United States, and of the peoples of sub-
Saharan Africa, and of African-Americans. Section 6 presents recent evidence
on the genetic basis of race differences in intelligence. Section 7 discusses the
geographical factors responsible for the evolution of racial differences and
Section 8 summarizes recent debates on this question.
The issue of geographical variation in intelligence has some degree of
overlap with that of race differences in intelligence. The reason for this is that
the human races evolved in different geographical locations, so that
geographical differences frequently amount to much the same thing as race
differences. For instance, the Negroid people evolved in sub-Saharan Africa
while the Mongoloid peoples evolved in North East Asia, so any differences in
intelligence between the two peoples is both a geographical and a racial
difference. The value of considering racial differences in intelligence from a
geographical standpoint is that it suggests an evolutionary and climatic theory
for the evolution of these differences, such as that presented in the last two
sections of this chapter.

2. H. J. EYSENCK'S WORK ON RACE AND INTELLIGENCE

Most of the initial work on race differences in intelligence was carried out on
blacks and whites in the United States. The first major study was carried out in
260 Intelligence

World War I on conscripted men. Intelligence test results for 10,936 white and
26,640 black men were calculated by Yerkes (1921), who found a 17 IQ point
difference in favor of whites. In subsequent decades several hundred studies
have been published which have in general confirmed this conclusion. These
have been summarized by Shuey (1966) and Osborne and McGurk (1982), who
concluded that the black-white differences have remained constant over a
period of approximately half a century. There have been three principal
controversies over this difference. These concern whether genetic factors are
operating; the precise nature of the difference and whether it lies in general
intelligence, spatial ability, verbal ability, etc.; and what were the factors
responsible for its evolution.
Eysenck's contribution to these questions lies in his endorsement of the
theory that genetic factors play a significant role in the black-white IQ
difference. However, he did not always take this view. He has recorded that in
the 1940s and 1950s he took the environmentalist view. In an autobiographical
sketch he writes that "I had no difficulty in telling my students that there was
no evidence for genetic differences in intelligence between racial groups, and
that the differences that existed were due entirely to environmental pressures
and disadvantages imposed on colored and other groups" (Eysenck, 1991, p.
17). During the 1960s, however, he began to have doubts about the
environmentalist position. He has recorded that these doubts were stren-
gthened in 1966 when he read Audrey Shuey's book The Testing of Negro
Intelligence (Shuey, 1966). Shuey's book was a compilation of several hundred
studies of the intelligence of blacks and whites in the United States which
consistently showed that blacks obtain average scores about 15 IQ points lower
than whites. She concluded that the evidence indicated the presence of genetic
factors as largely responsible for this difference. Eysenck thought about the
evidence and "I emerged with the firm impression that Shuey was right"
(p.18). This change in his thinking was confirmed in 1969 when Jensen (1969)
published his Harvard Education Review article in which he argued that genetic
factors make a significant contribution to the black-white difference in mean
IQ in the United States.
Eysenck took up this issue in his book Race, Intelligence and Education
(Eysenck, 1971). In this he argued in favor of the conclusions reached by Shuey
and Jensen. There were two principal considerations leading him to this view.
First, the evidence showed that when black and white children are matched for
schooling and parental socioeconomic status, there remains a 12 IQ point
difference between them. Second, the children of black middle class parents
obtain lower average IQs than the children of white lower class parents. This
shows that even when black children are reared in superior environments, as
indexed by higher socioeconomic status, they perform at a lower level than
white children reared in inferior environments. Eysenck concluded that only
the operation of genetic factors could provide convincing explanations of these
 Geographical variation in intelligence 261

findings. He wrote of Jensen's 1969 paper that "The facts, as we see them now,
support his tentative and carefully worded conclusions; all he finally concluded
was that there is a good deal of evidence (which he never considered
conclusive) to show that white-black IQ differences were in part genetically
caused" (p. 145).

3. GEOGRAPHICAL VARIATION IN INTELLIGENCE

In the 1960s and early 1970s the work of Shuey, Jensen, and Eysenck was
principally concerned with the issue of the difference in average intelligence
between blacks and whites in the United States. In the late 1970s I decided it
would be useful to examine race differences in a worldwide perspective,
particularly to see whether there are consistent differences in the intelligence
levels of the different races in a variety of geographical locations. This review
showed that the Caucasoid peoples of Europe, the United States, Australia, and
New Zealand obtain average IQs of around 100; there was a little evidence that
Oriental or Mongoloid peoples, mainly Chinese and Japanese, tend to score a
little higher; Negroid peoples invariably obtain low average IQs, not only in the
United States but also in Britain, the Caribbean, and in Africa; American
Indians in the United States score about midway between blacks and whites; and
Australian Aborigines obtain average IQs of around 80 (Lynn, 1978).
In the mid-1980s, I returned to this question and produced updated
worldwide literature surveys of the intelligence of the major races considered
in relation to the geographical and climatic environments in which they had
evolved (Lynn, 1987, 1991a, 1991b). These reviews summarized the research
evidence published up to the year 1990 and are summarized in Table 13.1. This
gives the geographical and climatic environments in which the seven major
racial groups evolved and their median IQs. As entered in this table, South
Asian Caucasoids consist of Indians from the Indian subcontinent. Negroid-
Caucasoid hybrids comprise blacks in the United States and Britain.
Australasians consist of Australian Aborigines, New Zealand Maoris, and
South Pacific islanders. Although the medians generally provide a good
estimate of the average intelligence level of the races, in the case of African
Negroids I proposed that the most satisfactory study in terms of sample size
and representativeness indicated that the mean IQ was approximately 70,
rather than the median of 75.
It was also noted in this survey that there are racial differences in the
patterns of abilities. The Mongoloid peoples have particularly strong spatial
ability, strong reasoning ability, but relatively weak verbal abilities. This pattern
is also present in the American Indians, although the whole profile is
depressed. Negroid peoples have the reverse pattern of particularly weak
spatial ability, weak reasoning ability, and relatively strong verbal ability.
262 Intelligence

Table 13.1. Median IQs of seven racial groups shown in relation to the geographical locations and
climates in which they evolved

Geographical
Race location Climate IQ
Mongoloids North-east Asia Cold temperate 105
European Caucasoids Europe Temperate 100
South Asian Caucasoids India Warm temperate 91
Amerindians America Temperate-tropical 90
Australasians Australia Temperate-tropical 90
Caucasoid-negroids USA-UK Temperate-tropical 84
Negroids Africa Tropical-subtropical 75

In this survey it was argued that the racial differences in intelligence test
scores were corroborated by three further kinds of evidence. First, there are
race differences in reaction times, showing fastest reaction times in Mongo-
loids, intermediate reaction times in Caucasoids, and slowest reaction times in
Negroids. Reaction times are widely regarded as measures of the neurological
efficiency of the brain and as the neurophysiological basis of intelligence (e.g.,
Eysenck, 1982; Jensen 1982; and see chapters 11, 12, and 14 in this volume).
Hence, the racial differences in reaction times demonstrate that the differences
obtained on intelligence tests cannot be ascribed to cultural factors, bias in the
tests and so on, but are present at the level of basic neurophysiological
functioning.
Second, it was argued that the race differences in the development of early
civilizations documented by Baker (1974) can be understood in terms of the
differences in intelligence. Baker showed that the Caucasoids in the Near East
and the Mongoloids in China were the only peoples who developed sophisticated
early civilizations which included the inventions of the wheel, money, writing,
arithmetic, metal working, animal drawn vehicles, agriculture, the domestication
of animals, legal systems, and the use of stone or brick for the building of houses
in cities. Baker showed that the American Indians achieved about half of these
attributes of civilization, while the Negroids achieved none of them.
Third, evidence was reviewed showing that brain size is positively associated
with intelligence and that there are race differences in average brain size,
Mongoloids having the largest average brain size and Negroids the smallest.
This provides further evidence that the race differences in intelligence have a
firm neurophysiological basis and cannot be attributed solely to cultural
differences.

4. THE INTELLIGENCE OF MONGOLOIDS IN EAST ASIA

We turn now to recent studies on the intelligence of the Mongoloid peoples in

east Asia and in the United States. The results of these are summarized in
 Geographical variation in intelligence 263

Table 13.2. This gives the geographical location of the samples, the number of
subjects tested, their age, and the mean IQ for general intelligence
(Spearman's g), verbal, and spatial abilities. The figures for general intelligence
are obtained either from tests of reasoning ability or from a combination of
scores for reasoning, verbal, and spatial abilities, whichever seems more
appropriate for the particular set of data. All the IQs are calculated in relation
to a mean of 100 and standard deviation of 15 for European or North
American Caucasoids. The details of the studies are explained and commented
upon in the text that follows.
The first study entered in Table 13.2 consists of Chinese IQs derived from
the Chinese standardization of the Wechsler Intelligence Scale for Children—
Revised (WISC-R) carried out by Li, Jin, Vandenberg, Zhu, and Tang (1990).
The standardization was carried out in 1984 on a representative sample of 660
children aged 6-16 in Shanghai (30 boys and 30 girls for each of the 11 age
groups). The Chinese standardization consists of all the same subtests as the
American test, the verbal items having been translated into Chinese
equivalents. To calculate Chinese IQs in relation to American, the Chinese
means for each subject for each age group have been converted to American
scaled scores and the IQs computed for each age group. This gives the Chinese
a full-scale IQ of 112.4, a verbal IQ of 119.7, and a Performance IQ of 105.1.
Some adjustments need to be made to these figures. First, the mean IQ of
American whites on the WISC-R is 102.2 (Jensen & Reynolds, 1982), so 2.2 IQ
points need to be subtracted to give Chinese IQs in relation to American white
Table 13.2. Mean IQs for various Mongoloid populations in relation to a white IQ of 100

Verbal Spatial
Location N Age Test IQ IQ Reference
China 660 6-16 WISC-R 107.2 115.5 114.2 Li et al. (1990)
China 297 14-15 various 103.0 - 104.6 Li, Sano and
Merwin (1996)
Japan 239 14-15 various 103.0 - 109.2 Li, Sano and
Merwin (1996)
Japan 455 5-7 CCAT 106.9 121.7 - Takeuchi and Scott
(1992)
South Korea 440 2-12 K-ABC 105.7 - - Moon (1988)
South Korea 107 9 various 109.0 98.0 111.0 Lynn and Ja Song
(1994)
Taiwan 43825 6-7 CPM 104.6 - - Hsu (1976)
Taiwan 2476 9-12 SPM 104.7 - - Lynn (1997)
United States - - various 103.5 97 110 Vernon (1982)
United States - - various 97.5 95.4 - Flynn (1991)
United States 167 9-11 SPM 105 - - Jensen and Whang
(1993)
United States 42 - AFQT 103 - - Herrnstein and
Murray (1994)
United States 48 6-17 DAS 104.4 100.5 104.7 Lynn (1996a, b)
264 Intelligence

IQs of 100. Second, the Chinese standardization was carried out 10 years later
than the American, and American IQs have been increasing on the Wechsler
tests at a rate of 3 IQ points per decade on the full-scale IQ, 2 IQ points per
decade on the Verbal, and 4 IQ points on the Performance IQ (Flynn, 1980).
These figures need to be subtracted from the Chinese IQs to secure
comparability as regards date of standardization. Third, the Wechsler tests
do not provide a clear measure of spatial ability. The only test of spatial ability
is the Block Design test. The mean American scaled score of the Chinese on
this is 12.4, equivalent to an IQ of 117.
The following adjustments have been made to the results of the Chinese
standardization's to reach the figures entered in Table 13.2 The figure of 107.2
for g is the full-scale IQ in relation to American whites and reduced for the
later date of the Chinese standardization; the figure of 115.5 for verbal ability is
the verbal IQ similarly adjusted; the figure of 114.2 for spatial ability is
calculated from the Block Design test, also adjusted downwards. The Chinese
means for general intelligence (IQ 107.2) and spatial ability (IQ 114.2) are
within the range of previous studies. The verbal IQ, however, is much higher
than has been found in other samples. The high Chinese verbal IQ is mainly
derived from their exceptionally high scores on the Arithmetic and Vocabulary
tests. Possibly the explanation for these is simply that arithmetic is taught with
exceptional thoroughness in Chinese schools and that the vocabulary test was
made too easy in the Chinese translation.
The second study shown in Table 13.2 is also for China. Li, Sano, and
Merwin (1996) tested 14-15-year-olds drawn as representative samples from
city schools in China, Japan, and the United States. The American sample (N
= 318) came from four public schools in Minneapolis and St. Paul; 95% were
white and the remaining 5% "other." The Chinese came from three schools in
and around Beijing and the Japanese were from four schools in Toyama city.
The testing was carried out in 1985. Six tests were given and the authors
present their data as raw score differences, for which they give the standard
deviations. To afford comparison with other studies, I have converted scores
into conventional IQs for Japan and China in relation to American IQs of 100
and standard deviation of 15. This has been done by calculating the American-
Japanese and the American-Chinese differences in standard deviation units,
using the combined standard deviation of all three samples, and converting
these to IQs.
The six tests used in the study and the IQs obtained by the Chinese and
Japanese were as follows: Figure classification, a test of nonverbal or abstract
reasoning ability, on which the Japanese obtained a mean IQ of 113.1 and the
Chinese a mean IQ of 109.8; Pedigree analysis, a test of verbal reasoning ability
("John is the uncle of Sarah, who is the aunt of Jill; what relation is Jill to
John?"), on which the Japanese obtained an IQ of 93.0 and the Chinese an IQ
of 96.3. The authors of the study comment that the Chinese and Japanese were
 Geographical variation in intelligence 265

handicapped on this test because family pedigrees are much more complex in
their languages than they are in English. For instance, there are 12 words for
"cousin" in Chinese depending on whether the individual concerned is male or
female, on the father's or the mother's side, and on his or her age. This means
that the results must be an underestimate of Chinese and Japanese IQs, but it is
not possible to adjust for this. The best way of treating the results would seem to
be to combine the two reasoning tests to provide a measure of g. This gives a
value of 103.0 for both the Chinese and the Japanese entered in Table 13.2.
The study used two tests of spatial ability. These were hidden pictures, a test
of spatial visualization on which the Chinese obtained a mean IQ of 111.4 and
the Japanese of 113.7, and card rotation, on which the Chinese obtained a
mean of 97.8 and the Japanese a mean of 104.8. The results of these two tests
can be averaged to produce IQs for spatial ability of 104.6 for the Chinese and
109.2 for the Japanese. The general pattern of the results confirms previous
studies showing that the Chinese and Japanese score particularly high on
spatial ability. It is interesting to note that the Chinese and Japanese both
perform better on the test of spatial visualization than on the spatial rotation.
These results make an important contribution by defining more precisely than
hitherto the nature of the Mongoloid strength on the spatial ability, which
evidently lies principally in spatial visualization.
The fifth test used in this study consisted of arithmetical reasoning. Both the
Chinese (IQ 126.7) and the Japanese (IQ 110.3) performed much better than
the Americans. This can be attributed partly to their high reasoning abilities
and also, particularly in the case of the very high scores of the Chinese, to
intensive teaching of math in schools.
The sixth test used in the study was designated number comparison and is a
test of perceptual speed. The Chinese performed worse than the Americans
(96.1) and the Japanese performed only fractionally better (101.2). These
results confirm the previous studies showing that Mongoloid peoples do not
have any significant advantage over Caucasoids on this ability.
The Li, Sano, and Merwin (1996) study provides an interesting comparison
with a previous study by Stevenson et al. (1985) which employed a similar
research design involving the testing of 240 6-year-olds and 240 11-year-olds in
the Japanese city of Sendai, with equal numbers in the Taiwanese capital,
Taipei, and in the American city of Minneapolis. The results were that there
were no differences between the Japanese, Taiwanese, and American children.
This is the only study in which it has not been found that Mongoloids obtain
higher average IQs than Caucasoids. The explanation for the anomalous
results may lie in the choice of Minneapolis as a typical white American city.
There is substantial evidence that the average intelligence level of whites in
Minneapolis is higher than that among American whites as a whole. This
evidence comes from military draft data for 1968 presented by Jensen (1973,
p.64). The percentage of drafted whites failing the Armed Forces Classification
266 Intelligence

Test ranged from a low of 0.6 in Rhode Island to a high of 9.7 in Tennessee.
The percentage failing in Minnesota, in which the city of Minneapolis is
situated, was 0.9% and was the second lowest in the whole of the United
States. A low percentage failure rate is indicative of a high average IQ.
Furthermore, Flynn (1980, p. 107) has calculated the mean IQ of whites in
Minnesota at 105 in relation to a mean of 100 for the whole of the United
States. The evidence is clear. If one wanted to select a city representative of the
United States for intelligence, one of the last to be chosen would be
Minneapolis. If we adopt Flynn's estimate that the IQ in Minneapolis is 105,
the finding that children in Minneapolis obtain the same average IQ as those in
Sendai and Taipei becomes consistent with the other studies indicating that
Mongoloid IQs are about 5 IQ points higher than those of Caucasoids. It is a
coincidence that Li et al. (1996) also selected Minneapolis as the American city
with which to compare the test scores of adolescents in Toyama and Beijing.
To adjust for the high IQ in Minneapolis, 5 IQ points should be added to the
advantage obtained by the Japanese and Chinese students, raising their general
IQs from 103 to 108.
The next study entered in Table 13.2 was carried out by Takeuchi and Scott
(1992) and comprised a comparison of the mean IQs obtained by 454 Japanese
5-7-year-old children in the city of Nagoya on the Canadian Cognitive Abilities
Test (CCAT). The test measures abstract reasoning (g), verbal comprehension,
and quantitative abilities. The Japanese children's abstract reasoning IQ of
106.9 is in the general range of results for Mongoloids. Their verbal IQ of 121.7
is exceptionally high and raises doubts about whether the vocabulary items
used in the test may have been rendered too easy for the Japanese children.
The translation problem makes comparisons of verbal abilities between
populations speaking different languages more error-prone than is the case
with nonverbal tests. On the quantitative scale of this test the Japanese
children obtained a mean IQ of 113.5. The Canadian Cognitive Abilities test
was standardized in Canada in the late 1980s and no allowance has been made
for the slightly later date of the Takenchi and Scott study.
My 1991 review of the literature of the intelligence of the Mongoloids did
not include any studies from South Korea. In the present review it has proved
possible to include two studies. The first of these was carried out by Moon and
consists of the administration of Kaufman's Assessment Battery for Children
(K-ABC) Test to 40 Korean children of each age from 2 through 12 years,
making a total of 440 children. The K-ABC provides a measure of
"simultaneous processing" which is a measure of nonverbal reasoning and in
the Korean sample correlated .73 with the Wechsler full-scale IQ. It can be
considered a measure of general intelligence. The mean IQ of the Korean
children was 107.2. The Korean children were tested approximately five years
 Geographical variation in intelligence 267

after the American children, so 1.5 IQ points have been subtracted from their
mean to equate them for the year of administration, giving them an IQ of
105.7.
A further study of intelligence in South Korea has been published by Lynn
and Song (1994). Representative samples of 107 Korean and 115 British 9-year-
olds were given four tests of different cognitive abilities, with the following
results. The Korean children had an advantage of 9 IQ points on the Progressive
Matrices, a test of nonverbal reasoning and Spearman's g; they had an advantage
of 11 IQ points on a test of spatial ability, and an advantage of 6.4 IQ points on a
test of perceptual speed. British children scored 2 IQ points higher on a test of
verbal fluency. The results confirm the typical pattern of high reasoning and
spatial abilities and lower verbal abilities among Mongoloids.
We turn next to studies of intelligence in Taiwan. Entered first in Table 13.2
is a study by Hsu (1976) which I failed to pick up in my 1991 review. This
investigation consisted of the administration of the Colored Progressive
Matrices Test (CPM) in 1975 to virtually all first grade children with a mean
age of 6.8 years in Taipei. Their mean score was 18.6. This is equivalent to a
mean of 18 on the British standardization sample of the Standard Progressive
Matrices and to the 57th percentile, and this percentile is equivalent to an IQ
of 102.7. The test was administered in Taiwan seven years before the
standardization of the CPM in Britain, with which the Taiwanese mean is being
compared. Mean scores on the CPM in Britain have increased by 2.7 IQ points
per decade over the period 1969-1982 (Lynn & Hampson, 1986), so to adjust
for the time difference between the Taiwan and British means we need to add
1.9 IQ points to the Taiwanese mean, bringing it up to 104.6.
The second study of intelligence in Taiwan comes from the Taiwanese
standardization of the Standard Progressive Matrices, which was carried out on
a large sample of 9-12-year-olds in 1989. To calculate the Taiwanese IQ, I have
converted the mean scores obtained by the Taiwanese children on the test to
British percentiles and transformed these percentiles into IQs (Lynn, 1997).
This gives the Taiwanese an IQ of 106.8. This figure has been adjusted
downwards by 1.9 IQ points to allow for the British standardization having
been carried out 10 years earlier than the Taiwanese, this being the rate of
secular increase of the scores of British children on the Progressive Matrices
(Lynn & Hampson, 1986). This gives the Taiwanese a mean IQ of 104.7.
Taking an overall view of the eight new studies of the intelligence of
Mongoloid populations in East Asia, it is evident that they confirm the results
of the earlier studies for which the median IQ was 105. The eight new studies
show a mean IQ in the range of 103.0 and 109.0, with a median of 105.2. The
results of the earlier studies were criticized by Brody (1992) on the grounds
that there may have been differences in the administration of the tests or in the
sampling which could account for the higher IQs obtained by Mongoloids.
Even at the time this criticism was made, it is difficult to believe that these
268 Intelligence

biases could have affected all 25 studies in the same direction. We now have a
further eight studies, all of which show a mean IQ for Mongoloids in the range
of 103-109. Brody's suggestion that biases could have consistently favored the
results obtained by Mongoloids in 33 studies strains credulity beyond breaking
point.

5. INTELLIGENCE OF AMERICAN ETHNIC MONGOLOIDS

Recent studies of the intelligence of American ethnic Mongoloids are

summarized in the bottom rows of Table 13.2. The starting point for the
consideration of the intelligence of American ethnic Mongoloids (i.e., ethnic
Chinese, Japanese, Koreans, and some Vietnamese) has its roots in a survey of
the literature published in the early 1980s by Vernon (1982). Vernon
concluded that American ethnic Mongoloids have an average IQ of around
110 on nonverbal and spatial tests and 97 on verbal tests. If these are averaged
to 103.5, the thesis of the high Mongoloid IQ is confirmed. However, this
conclusion has been challenged by Flynn (1991) who argues that Vernon failed
to take account of outdated test norms and that after adjustment, these
American ethnic Mongoloids have a mean nonverbal IQ of 99.6 and a mean
verbal IQ of 95.4, which can be averaged to 97.5. There are, however, a number
of problems with Flynn's analysis which I have discussed at length in Lynn
(1993b). The two chief problems are that there has never been a sound study of
a representative sample of American ethnic Orientals, and both Vernon's and
Flynn's conclusions are built up from poor serendipity samples; and that
Flynn's computation omits spatial ability on which Mongoloid peoples
invariably score high. Flynn's "nonverbal" IQ is not spatial ability but abstract
reasoning ability, as measured by tests like the Progressive Matrices.
Since 1991, three studies have appeared which provide better data on the
intelligence of American ethnic Mongoloids. The first of these comes from
Jensen and Whang (1993) who administered the Standard Progressive
Matrices to a sample of 167 9-11-year-old ethnic Chinese children in
California. Their average age was 10.4 and their mean score 41.9. This
corresponds to the 69th percentile of the British 1979 standardization sample,
and this converts to an IQ of 107.5. Assuming the data were collected in 1991,
we need to subtract 2.3 IQ points to allow for the rise in British norms over the
12 year period 1979-1981, reducing the IQ of the sample to 105.2. Jensen and
Whang adopt a different procedure for estimating the IQ of the American
ethnic Chinese sample. This was to test a comparison sample of 585 white
children and calculate the difference between the Chinese and the whites in
standard deviation units. The result of their calculations was that the Chinese
had a 5 IQ point advantage. Notice that the two methods of estimation
produce virtually identical results.
 Geographical variation in intelligence 269

The second recent study of the intelligence of American ethnic Mongoloids

is Herrnstein and Murray's (1994) analysis of the data of the National
Longitudinal Study of Youth, given the Armed Forces Qualification Test
(AFQT). From these data they compute mean IQs of 85 for blacks, 89 for
Hispanics, 103 for whites, and 106 for East Asians (Chinese, Japanese, and
Koreans). Thus the mean IQ of this sample of Mongoloids is 3 IQ points above
that of American Caucasoids. The third recent study consists of an analysis of
the data of the American standardization of the Differential Ability Scale. The
standardization sample gives mean IQs for American Asians for four abilities.
In relation to means of 100 for whites, American Asians obtained means of
104.4 for "general cognitive ability" (general IQ), 106.5 on nonverbal
reasoning, 104.7 on spatial ability, and 100.5 on verbal ability (Lynn, 1996).
Asian Americans are not precisely the same as Mongoloids because they
include non-Mongoloid Asians but the 1980 census shows that approximately
two-thirds of them are ethnic Chinese, Japanese, Koreans, and Vietnamese,
who can be considered as Mongoloids. The remaining one third of non-
Mongoloid American Asians are unlikely to contaminate the results seriously.
Notice that the results confirm the previous data indicating not only the high
overall IQ of Mongoloid peoples but also their pattern of higher reasoning and
spatial than verbal abilities.
These three studies appearing in the years 1991-1996 have gone some way
towards resolving the issue of the intelligence of American ethnic Mongoloids.
They have shown that it is 3-5 IQ points higher than that of Caucasoids, and is
therefore consistent with the numerous studies of the intelligence of
Mongoloid populations in the countries of the Pacific Rim. Furthermore, the
same pattern of strong spatial and weaker verbal intelligence is present in
American ethnic Mongoloids as has been found in Pacific Rim Mongoloids,
suggesting a genetic basis to this ability profile.

6. INTELLIGENCE IN SUB-SAHARAN AFRICA

Recent studies of intelligence in sub-Saharan Africa are summarized in Table

13.3. In the first of these, Owen (1992) administered the Standard Progressive
Matrices to 1056 white, 1093 black, 778 colored and 1063 Indian school
students aged approximately 16 years in South Africa. Owen presents the mean
scores and standard deviations for the four groups. He also gives the
differences between the whites and the other three groups in standard
deviation units, which can be easily converted into standard IQ differences.
The results are that the blacks scored 2.78 standard deviation units below the
whites, equivalent to an IQ of 58 in relation to a white IQ of 100; the coloreds
scored 1.35 standard deviations below the whites, equivalent to an IQ of 80;
270 Intelligence

Table 13_3. Intelligence in sub-Saharan Africa in relation to a white IQ of 100

Verbal Spatial
Location N Age Test g IQ IQ Reference
South Africa 1093 16 SPM 58 Owen (1992)
(Negroids)
South Africa 778 16 SPM 80 Owen (1992)
(Coloreds)
South Africa 1063 16 SPM 92.5 Owen (1992)
(Indians)
Zimbabwee 204 12-14 WISC-R 67.1 69.7 69.8 Zindi (1994)
Zimbabwe 204 12-14 SPM 72.4 Zindi (1994)
Ethiopia 250 15-16 SPM 69 Lynn (1994a)
Kenya 205 adults CPM 74 Boissiere,Knight, &
Sabot (1985)
Tanzania 179 adults CPM 69 - - Boissiere, Knight &
Sabot (1985)
Ghana 1639 15 CPM 60 - - Glewwe & Jacoby
(1992)

and the Indians scored 0.52 standard deviations below the whites, equivalent to
an IQ of 92.5.
The results contain three points of interest. First, the mean IQ of 55 of
blacks in this study is even lower than that normally found in studies of blacks
in sub-Saharan Africa. The explanation for this low figures lies in the
methodology for estimating the black-white difference which differs from that
which I have usually employed. Owen's study compares whites and blacks in
South Africa rather than calculating the IQ of blacks in relation to the
standardization data of British whites. If we adopt my usual method, we find
that the mean raw score of the blacks was 27.6. This corresponds to the third
percentile of the British standardization sample, which is equivalent to an IQ
of 72. Thus calculated in this way, the study provides a close confirmation of
many previous studies of the sub-Saharan IQ. The reason that Owen's study
yields a lower figure lies in the rather small standard deviation of the white
sample, the effect of which is to increase the black-white difference measured
in standard deviation units. In broad terms, Owen's study clearly provides
confirmation of the conclusion that the mean IQ of sub-Saharan blacks is
around 70 or maybe rather lower.
The second interesting result in Owen's studies concerns the mean IQ of 80
obtained by his sample of coloreds. These are a hybrid population of black and
white ancestry, and the striking result is that their average IQ stands almost
exactly midway between the white IQ of 100 and the black IQ of 58. This would
be expected from a genetic theory of the difference, because a hybrid
population should score intermediate between the two parent races. This
confirms the interpretation I proposed for the mean IQ of approximately 85 of
American blacks being higher than that of African blacks, namely that
 Geographical variation in intelligence 271

American blacks are racial hybrids with a significant amount of white ancestry.
These results contribute to the somewhat inconclusive debate over whether the
intelligence of American blacks is related to their proportion of Caucasian
genes. The issue is difficult to resolve among the American black population
because Caucasian genes for intelligence and for skin color or other markers of
Caucasian ancestry are likely to segregate independently after several
generations, with the result that no relationship between them would be
expected. Owen's South African data provide a clear verification of the genetic
theory that black-white hybrids (coloreds) will obtain IQs about midway
between blacks and whites.
The third interesting feature of Owen's study is the average IQ of 92.5
obtained by the Indians. The South African Indians are largely the descendants
of Indians brought to South Africa in the nineteenth century to work on the
cotton plantations. Their average IQ today is very much higher than that of the
blacks and coloreds and approaches that of the whites. This confirms studies of
the intelligence of Indians in Britain, whose average IQ is about 96. When
Mackintosh and Mascie-Taylor (1985) produced this result for British Indians,
they were unable to offer any explanation for why British Indians obtained a
significantly higher average IQ than British blacks. Yet considered in
evolutionary terms, the solution is obvious: Indians are Caucasians and would
be expected to obtain IQs closely similar to those of European Caucasians.
Another recent study of intelligence in sub-Saharan Africa has been
published by Zindi (1994), a Zimbabwean at the University of Zimbabwe. He
tested a sample of 204 black secondary school pupils aged 12-14 years with the
WISC-R (Wechsler Intelligence Scale for Children—Revised) and with the
Progressive Matrices (he does not state whether these are the colored or the
standard form of the test). Zindi gives the mean WISC-R of his sample as 67.1,
the mean verbal IQ as 69.7, and the mean performance IQ as 69.8. The last
figure has been entered in Table 13.3 in the spatial IQ column, although it
should be noted that the Wechsler performance scale is a poor measure of
spatial ability. Zindi gives the mean IQ of his sample on the Progressive
Matrices as 72.4, although he does not describe how he calculated this figure.
The third recent study of the black African IQ comes from Ethiopia. Kaniel
and Fisherman (1991) reported the results of 250 15-16-year-old Ethiopian
Jewish immigrants into Israel tested with the Standard Progressive Matrices. I
have calculated that the IQ of these Ethiopians is 69 (Lynn, 1994a).
The last three entries in Table 13.3 consist of three studies in which the
Colored Progressive Matrices (CPM) were administered to samples in Kenya,
Tanzania, and Ghana. The first two of these studies appeared in 1985 but I
failed to pick them up in my 1991 review, the reason for this being that they
were published in the American Economic Review, an unusual outlet for work
on the level of cognitive ability in Africa. The samples were adults and came
from Nairobi and Dar es Salaam. The mean score of the Nairobi (Kenya)
272 Intelligence

sample was 27.8. This is equivalent to a score of 32 on the Standard Progressive

Matrices (J. C. Raven, Court, & J. Raven, 1995), to the 4th percentile of adults
on the British 1979 norms, and a British IQ of 74. The mean score of the Dar
es Salaam (Tanzania) sample was 26.4, equivalent to a score of 29.5 on the
Standard Progressive Matrices, to the 2nd percentile on the British 1979
norms, and a British IQ of 69.
The final study is of 1639 adolescents in Ghana drawn as a random sample
from the entire country, reported by Glewwe and Jacoby (1992). Their mean
age was 15.2 and their mean score 12.5. This is equivalent to 11.5 on the
Standard Progressive Matrices, to the zero percentile of British 15 year olds on
the British 1979 norms, and to a British IQ of 60. Looking at the results as a
whole for intelligence in sub-Saharan Africa, the seven studies of Negroids
produce mean IQs in the range of 58 to 74, with a median of 69. This compares
closely with the mean IQ of 70 estimated for these populations in my 1991
review.

7. THE GENETIC BASIS OF RACE DIFFERENCES IN INTELLIGENCE

The issue of whether there is a genetic basis to differences in intelligence had

continued to be debated in the 1990s. The view that genetic differences are
involved has been strengthened by three new sources of evidence. These are
the consistency of the differences across the world, the effects of the adoption
of black babies by whites, and the race differences in brain size. The
consistency of the race differences across the world has been amply confirmed
by the studies reviewed in this chapter which have shown that Mongoloid
populations invariably obtain mean IQs a little above Caucasoids, while
Negroid IQs are much lower. This is what would be expected if the race
differences are substantially determined by genetic factors. It would not be
expected on an environmental theory. Environmentalists typically explain race
differences in terms of poverty (e.g., Brooks-Gunn, Klebanov, & Duncan,
1996) and this is superficially persuasive in the United States because poverty is
related to low intelligence and blacks are poorer than whites. The global
evidence runs counter to this thesis. The most striking case is China, which is a
very poor country because its economic development has been impeded by
communism, but the poverty in China has not depressed the intelligence of the
population. The intelligence of the Chinese is much higher than that of
American blacks, although they are much poorer.
A similar instance to appear recently concerns the level of intelligence in
Bulgaria. A standardization of the American Culture Fair Test carried out in
1982 shows that the mean IQ of Bulgarian children is 95 (Lynn, Paspalanova,
Stefinsky, & Tzenova, 1997). Yet these Bulgarian children were much poorer
than American blacks. The mean per capita income in Bulgaria at the time of
 Geographical variation in intelligence 273

the study was approximately 20% of that in the United States. But after tax per
capita incomes of American blacks, assessed by the US Bureau of the Census
(1986), at this time were 60% of those of American whites. Thus, although the
per capita incomes in Bulgaria were about one third of those of American
blacks, their intelligence levels were substantially higher and close to those of
American whites. The results suggests that race is a much more important
determinant of intelligence than incomes.
A second recent study on this issue addresses the question of whether the
black-white IQ difference in the United States has been narrowing since the
1960s in parallel with the improvement in the socioeconomic position of blacks.
This thesis was first proposed by Vincent (1991) and has been endorsed by
Herrnstein and Murray (1994) and by Neisser et al. (1996). Yet a study of the
black-white difference in 1986 showed that for 6-17-year-olds it was 13.5 IQ
points and for 2-6-year-olds, 15.1 IQ points, closely similar to the difference of
around 15 IQ points that has been present since 1917-1918 (Lynn, 1996). The
evidence suggests that the higher average IQ of American whites, relative to
that of blacks, remains unaffected by the improvement of the social and
economic position of blacks since the 1960s.
Further recent data supporting the case for a genetic basis for race
differences in intelligence consists of the results of the transracial adoption
study of black babies adopted by white graduate parents, carried out by
Weinberg, Scarr, and Waldman (1992). For a number of years it has been
argued that the decisive way to test the genetic hypothesis of the black-white
intelligence gap would be to examine the IQs of black babies reared in a white
environment. Thus, Rose, Kamin, and Lewontin (1984, p. 127) have written
that "the only way to answer the question of genetic differences in IQ between
groups would be to study adoption across racial and class boundaries." This is
precisely what Weinberg, Scarr, and Waldman did and their results on adopted
black children, tested at the age of approximately 17 years, were that their
intelligence levels remained the same as those of blacks reared by their own
parents. I have argued this conclusion in Lynn (1994b). The authors of the
study initially attempted to dispute my interpretation of it (Waldman,
Weinberg, & Scarr, 1994), but more recently Scarr has conceded that my
interpretation is correct, because she writes that "those adoptees with two
African American birth parents had IQs that were not notably higher than the
IQ scores of black youngsters reared in black families" (Scarr, 1995, p.7). Thus,
the crucial study has now been carried out and has shown that blacks reared in
a white environment fail to register any gains in intelligence. The result can
only be interpreted as providing strong evidence for genetic determination of
the black-white difference.
The third recent set of data substantiating a genetic basis for racial
differences in intelligence consists of race differences in brain size. There is
unquestionably an association between brain size and intelligence, shown in a
274 Intelligence

historical review of the evidence published in 1990 (Lynn, 1990a) and

confirmed in updated reviews by Jensen and Sinha (1993) and Rushton (1995).
There has also been an accumulation of recent data demonstrating average
racial differences in brain size such that brain size is greatest in Mongoloids,
intermediate in Caucasoids, and smallest in Negroids. The evidence indicates
that the average brain size of whites exceeds that of blacks by approximately
4%, while the average Mongoloid brain is slightly larger than that of
Caucasoids. These brain size differences are present both in absolute terms
and when body size differences are controlled. The evidence on body size
shows that blacks have about the same body size as whites, indicating that their
physical development is not retarded by poverty (Lynn, 1990b; 1993a; Rushton,
1995).
In spite of these results, the genetic basis of race differences in intelligence
continues to be disputed. In 1995 the American Psychological Association set
up a Task Force under the chairmanship of Urich Neisser to report on the
current state of knowledge on intelligence. On the issue of a possible genetic
basis to race differences in intelligence, the Task Force concluded that "There
is not much direct evidence on this point, but what little there is fails to support
the genetic hypothesis" (Neisser, 1996, p. 83). The authors of the report do not
say what they mean by "direct evidence," but it is difficult to see what evidence
could be more direct than the worldwide consistency of race differences in
intelligence, the negative results of the transracial adoption study, and the well-
documented existence of race differences in brain size. None of these were
mentioned in the Task Force report. If the authors of the report had taken a
closer look at the evidence they could not have failed to reach the conclusion
that the case for some genetic basis for race differences in intelligence can no
longer be disputed.

8. THE EVOLUTION OF RACE DIFFERENCES IN INTELLIGENCE

The genetically based racial differences in intelligence must have arisen

because the races evolved in different geographical locations, some of which
exerted stronger selection pressures for an increase of intelligence than others.
The Caucasoid and Mongoloid peoples evolved in Europe and Asia, while the
Negroid peoples evolved in Africa, so there must have been something about
the conditions in Europe and Asia which exerted stronger selection pressure
for an increase of intelligence than was present in Africa. To explain the
slightly higher intelligence levels of the Mongoloids than the Caucasoids, these
selection pressures for enhanced intelligence must have been a little stronger
in North East Asia than they were in Europe.
 Geographical variation in intelligence 275

A coherent theory of racial differences in intelligence has to provide an

account of what these different selection pressures were. Genetic theorists
have been criticized for a failure to do this. For instance, in an attack on The
Bell Curve, Easterbrook (1995, p. 36) writes that "Neither Herrnstein and
Murray nor any other credentialed believer in the brain-gene theory has
suggested how, on an evolutionary basis, black and white intelligence DNA
could have diverged significantly."
This assertion betrays an ignorance of the research literature characteristic
of critics of the genetic theory. In the last decade two theories have been
proposed to explain the evolutionary processes through which race differences
in intelligence have evolved. These are my cold winters theory and Rushton's
climatic predictability theory. I first proposed the cold winters theory in 1987 to
explain the high intelligence of the Mongoloids (Lynn, 1987) and later
extended the theory to all the human races (Lynn, 1987, 1991b). The theory
states that humans first evolved in the tropical and subtropical environments of
central Africa, where survival is relatively easy because plant and insect foods
are available throughout the year. Some of these peoples migrated northwards
into Asia and Europe, where they evolved into the Mongoloids and
Caucasoids. These peoples encountered a more hostile environment, the
crucial feature of which was the cold winters. These posed two problems. First,
there were no plant foods available during the winter months. The result of this
was that for a substantial part of the year, the Mongoloid and Caucasoid
peoples became wholly dependent on meat foods. This meant that they had to
develop the skills required for hunting large animals. Second, to survive during
the cold winters, these peoples had to learn how to build shelters, fabricate
clothing, and make fires.
All these problems were cognitively demanding, particularly during the ice
ages, when Eurasia was significantly colder than it is today. Survival during
these conditions required an increase in intelligence in the Caucasoids and
Mongoloids. The severest environment was that of North East Asia, which had
significantly colder winters than Europe, and this explains why the Mongoloids
have evolved the highest intelligence level.
The cognitive demands of survival in these hostile environments which have
acted particularly strongly on the spatial abilities which are necessary to plan
and execute effective group hunting of large swift-footed mammals, and to
make the weapons and tools required for killing and butchering them. This
explains why the Mongoloid peoples have evolved their high spatial abilities.
The climate in Europe was only marginally less severe than that in North
East Asia, explaining why the European Caucasoids evolved an intelligence
level closely approaching that of the Mongoloids. The environment in much of
South Asia was somewhat more benign, but here too there were cold winters,
explaining the high intelligence of the Indian Caucasoids.
276 Intelligence

The American Indians and the Australasians experienced more moderate

selection pressures for the enhancement of intelligence. The American Indians
are descended from an archaic Mongoloid peoples who inhabited North East
Asia around 50,000-60,000 years ago and migrated into the Americas around
40,000 years ago. During their sojourn in North East Asia these archaic
Mongoloids would have experienced the selection pressure of cold winters and
this would have driven up their intelligence. However, they escaped the full
rigor of the cold northern environment by migrating into the Americas before
the onset of the last ice age, which lasted between around 24,000 and 10,000
years. It appears that the experience of several thousand years in North East
Asia raised the intelligence of these archaic Mongoloids to an IQ of around 90
and that it stabilized at this figure after they crossed into the Americas. The
extreme cold of the last ice age must have been the selection pressure boosting
the intelligence level of the Mongoloids by a further 15 IQ points or so, and of
the European Caucasoids by a further 10 IQ points. This explains why the first
Mongoloid and Caucasoid civilizations were developed after the end of the last
ice age and not during any of the warm interludes which have occurred in
Eurasia during the course of the last 100,000 years. It was only after the
intelligence-boosting impact of the last ice age that the Mongoloids and
Caucasoids had become sufficiently intelligent to build civilizations.
What about the Pacific Islanders and Australasians? The ancestors of these
peoples must have experienced some exposure to cold winters during their
migration across southern Asia, but they found benign subtropical environ-
ments when they settled in the Pacific islands and northern Australia. The
result of this has been that during their evolutionary history they were
subjected to some selection pressure for enhanced intelligence, raising it above
that of the Negroids but below that of the other major races.
This geographical-evolutionary theory also explains why there are racial
differences in spatial ability in addition to those for general reasoning abilities.
Spatial ability is required to solve several of the problems encountered by the
Mongoloids and the Caucasoids, especially the construction of tools and
weapons and the development of group hunting techniques. This would have
provided the selection pressure leading to the enhancement of spatial ability
and explains why, over and above general intelligence, spatial ability is
strongest among the Mongoloids, intermediate in the Caucasoids, and weakest
in the Negroids (Lynn, 1991). The demand of spatial ability for hunting and
tool making also explains why this ability is stronger in males than in females.
Males specialized in the making of tools, weapons, and in hunting, while
females specialized in child rearing and gathering food plants, so males were
more subjected to selection pressure for the enhancement of spatial ability.
 Geographical variation in intelligence 217

9. RECENT DEBATES ON THE EVOLUTION OF RACE DIFFERENCES IN

INTELLIGENCE

The cold winters theory of the evolution of enhanced intelligence in the

Mongoloids and Caucasoids has been taken up and elaborated by Edward
Miller. His first contribution was to propose that because many foods were
only available seasonally for short periods in the cold environments, people
would have had to figure out how these could be stored for future consumption
and this would have been a cognitively demanding problem requiring an
increase of intelligence (Miller, 1991). He does not give examples, but one of
these would have been the exploitation of a salmon run. Salmon migrate up
rivers for a short period every year and during this time it is possible to catch a
lot of them by the use of spears or traps. Too many of them could have been
caught for immediate eating, so it would have been useful to preserve many of
these fish for future consumption. This could have been done by building ice
houses as a forerunner of the modern deep freeze or by splicing and drying
them. Fruits can also be preserved if they are kept in the right temperature and
conditions. Miller also suggested that the storage of food would have required
the capacity to think ahead and that this is an aspect of high intelligence.
Finally, he suggests that arithmetical skills would have been required for
allocating stored food equitably among members of the group.
In a series of later papers, Miller (1993, 1994, 1995) has developed his
"paternal provisioning theory" as a further component of the selection
pressures for increased intelligence in the Mongoloid and Caucasoid peoples.
His theory is that in tropical and subtropical Africa, Negroid females and
children had relatively little need for males to supply them with food. They
could gather their own food because in Africa plant and insect foods were
available throughout the year and are easily obtained by females and children.
In the colder environments of Eurasia there were no plant or insect foods for
large parts of the year, and females and children became wholly dependent on
males to supply or "provision" them with food, which they did hunting large
animals. The major impact of the necessity of paternal provisioning was that it
made males and females more dependent on one another in the Mongoloid
and Caucasoid peoples. Females depended on their males to feed them, and
males depended on the females they were provisioning to bear and rear their
children. This promoted the formation of stable male-female bonding among
Mongoloids and Caucasoids more strongly than among Negroids. Miller
suggests that the necessity for male provisioning of females in the northern
environments had a cognitive impact. This was that as males were making
considerable investment of energy in provisioning their females, they needed to
ensure that these females were faithful. This required an increase of
intelligence to detect female infidelity. Conversely, females had to develop
enhanced intelligence in order to conceal their infidelities from the males who
278 Intelligence

were provisioning them. Miller writes that "detecting deception in a mate calls
for a high level of reasoning" (1995, p.915). He has made some useful
additions to the list of cognitive demands imposed on Mongoloids and
Caucasoids by the cold environments of Eurasia.
An alternative to the cold winters theory of the evolution of race differences
in intelligence has been advanced by Rushton, who has proposed that the
crucial feature of the arctic and subarctic environments in which the
Mongoloids and Caucasoids evolved during the ice ages was that they were
more predictable than the tropical and subtropical environments of Africa
inhabited by the Negroids (Rushton, 1991, 1995; Rushton & Ankney, 1993).
This theory was proposed in the context of his r-K theory of race differences,
which states that there is a Mongoloid-Caucasoid-Negroid gradient in repro-
ductive strategies, such that the Mongoloids are the most K (small numbers of
children with high parental investment) and Negroids the most r (large
numbers of children and low parental investment). Rushton maintains that
high intelligence is associated with the K strategy. The general biological
principle governing the evolution of r and K strategies is that r strategies tend
to be adopted in unpredictable environments, and K strategies in environments
that are predictable. Rushton applied this principle to his theory of human race
differences.
Rushton's theory that it was the high predictability of the arctic and subarctic
environments that was responsible for the evolution of higher intelligence and
other characteristics in Mongoloids and Caucasoids has been criticized on
several grounds by Miller (1993, 1995). He agrees that the African Savanna
was an unpredictable environment but argues that this should have selected the
Negroids for higher intelligence because of the need to cope with un-
predictable occasional droughts. Miller concludes that the cold environment of
Eurasia provides a better explanation of the higher intelligence of the
Mongoloids and Caucasoids than its greater predictability. In a second paper
Miller elaborated this position and concluded that "the critical problem for
prehistoric people in the frigid North was to survive the winter" (1995, p. 912).
Rushton (1995) has responded to these criticisms by adding the cold winters
theory to his predictability theory to explain the evolution of high intelligence
in the Mongoloids and Caucasoids.

10. CONCLUSION

It is now approximately a quarter of a century since Hans Eysenck reached the

conclusion that the black-white difference in intelligence has a significant
genetic basis. Subsequent research has considerably strengthened this
conclusion, especially the mounting evidence for consistent racial differences
in intelligence in a wide variety of geographical locations, the results of the
 Geographical variation in intelligence 279

transracial adoption study, and the evidence for racial differences in average
brain size. The formulation of a coherent theory to explain the evolution of
these racial differences in intelligence as resulting from the selection pressure
of the cognitive demands of having to survive in cold geographical
environments, exerted on the Mongoloid and Caucasoid peoples, adds further
strength to the genetic theory. A good scientist is able to reach the correct
conclusion on the basis of incomplete evidence and this is what Hans Eysenck
was able to do in his 1971 book on the issue of racial differences in intelligence.

REFERENCES

Baker, J. (1974). Race. Oxford: Oxford University Press.

Boissiere, M, Knight, J. B., & Sabot, R. H. (1985) Earnings, schooling, ability and
cognitive skills. American Economic Review, 75, 1016-1030.
Brody, W. (1992). Intelligence (2nd ed.). New York: Academic Press.
Brooks-Gunn, J., Klebanov, P. K., & Duncan, G. J. (1996) Ethnic differences in
children's intelligence test scores: role of economic deprivation, home environment
and maternal characteristics. Child Development, 67, 396-304.
Easterbrook, G. (1995) Blacktop basketball and The Bell Curve. In R. Jacoby and N.
Glauberman (Eds.), The Bell Curve debate. New York: Times Books.
Eysenck, H. J. (1971). Race, intelligence and education. London: Maurice Temple Smith.
Eysenck, H. J. (1982). Introduction. In H. J. Eysenck (Ed.), A model for intelligence.
Berlin: Springer.
Eysenck, H. J. (1991). Introduction: science and racism. In R. Pearson (Ed.), Race,
intelligence and bias in academe. Washington, DC: Scott-Townsend.
Flynn, J. R. (1980). Race, IQ and Jensen. London: Routledge & Kegan Paul.
Flynn, J. (1991). Achievement beyond IQ. Hillsdale, NJ: Lawrence Erlbaum.
Glewwe, P., & Jacoby, H. (1992). Estimating the determinants of cognitive achievement in
low income countries. Washington, DC: World Bank.
Herrnstein, R. J., & Murray, C. (1994). The Bell Curve. New York: Free Press.
Hsu, C. C. (1976). The learning potential of first graders in Taipei city as measured by
Raven's Colored Progressive Matrices. Acta Pediatrica Sinica, 17, 262-274.
Jensen, A. R. (1969). How much can we boost IQ and scholastic achievement? Harvard
Educational Review, 39, 1-123.
Jensen, A. R. (1973). Educability and group differences. London: Methuen.
Jensen, A. R. (1982). Reaction time and psychometric g. In H. J. Eysenck (Ed.), A
model for intelligence. Berlin: Springer.
Jensen, A. R., & Reynolds, C. R. (1982). Race, social class and ability patterns on the
WISC-R. Personality and Individual Differences, 3, 423^38.
Jensen, A. R., & Sinha, S. N. (1993). Physical correlates of human intelligence. In P.
Vernon (Ed.), Biological approaches to the study of human intelligence. New York:
Academic Press.
Jensen, A. R., & Whang, P. A. (1993). Reaction times and intelligence: a comparison of
Chinese-American and Anglo-American children. Journal of Biosocial Science, 25,
397-410.
280 Intelligence

Kaniel, S., & Fisherman, S. (1991). Level of performance and distribution of errors in
the Progressive Matrices Test: a comparison of Ethiopian immigrant and native
Israeli adolescents. International Journal of'Psychology', 26, 25-33.
Li, D., Jin, Y., Vandenberg, S. G., Zhu, Y., & Tang, C. (1990). Report on Shanghai
norms for the Chinese translation of the Wechsler Intelligence Scale for Children—
Revised. Psychological Reports, 67, 531-541.
Li, X., Sano, H., & Merwin, J. C. (1996). Perception and reasoning abilities among
American, Japanese and Chinese adolescents. Journal of Adolescent Research, 11,
173-193.
Lynn, R. (1978). Ethnic and racial differences in intelligence: International compari-
sons. In R. T. Osborne, C. E. Noble, & N. Weyl (Eds.), Human Variation. New York:
Academic Press.
Lynn, R. (1987). The intelligence of the Mongoloids: a psychometric, evolutionary and
neurological theory. Personality and Individual Differences, 8, 813-844.
Lynn, R. (1990a). The role of nutrition in secular increases in intelligence. Personality
and Individual Differences, 11, 273-285.
Lynn, R. (1990b). New evidence on brain size and intelligence: a comment on Rushton
and Cain and Vanderwolf. Personality and Individual Differences, 11, 795-797.
Lynn, R. (1991a). Race differences hi intelligence: A global perspective. Mankind
Quarterly, 31, 255-296.
Lynn, R. (1991b). The evolution of racial differences in intelligence. Mankind Quarterly,
32, 99-173.
Lynn, R. (1993a). Further evidence for the existence of race and sex differences in
cranial capacity. Social Behavior and Personality, 21, 89-91.
Lynn, R. (1993b). Oriental Americans: Their IQ, educational attainment, and
socioeconomic status. Personality and Individual Differences, 15, 237-242.
Lynn, R. (1994a). The intelligence of Ethiopian immigrant and Israeli adolescents.
International Journal of Psychology, 29, 55-56.
Lynn, R. (1994b). Some reinterpretations of the Minnesota Transracial Adoption
Study. Intelligence, 19, 21-27.
Lynn, R. (1996a). Racial and ethnic differences in intelligence in the United States on
the Differential Ability Scale. Personality and Individual Differences, 20, 271-273.
Lynn, R. (1996b). Intelligence in Taiwan. Personality and Individual Differences, 22,585-
586.
Lynn, R. (in press). Intelligence in Taiwan. Personality and Individual Differences.
Lynn, R., & Hampson, S. (1986). The rise of national intelligence: evidence from
Britain, Japan and the United States. Personality and Individual Differences, 7,23-32.
Lynn, R. & Ja Song, M. (1994). General intelligence, visuospatial and verbal abilities of
Korean children. Personality and Individual Differences, 16, 363-364.
Lynn, R., Paspalanova, E., Stetinsky, D., & Tzenova, B. (in press). Intelligence in
Bulgaria. Social Behavior and Personality.
Lynn, R., & Song. M. J. (1993). General intelligence, visuospatial and verbal abilities in
Korean children. Personality and Individual Differences, 16, 363— 364.
Mackintosh, N. J., & Mascie-Taylor, C. G. N. (1985). The IQ question: In Education for
all, Cmnd paper 4453. London: HMSO.
Miller, E. M. (1991). Climate and intelligence. Mankind Quarterly, 32, 127-132.
 Geographical variation in intelligence 281

Miller, E. M. (1993). Could r selection account for the African personality and life
cycle? Personality and Individual Differences, 15, 665-676.
Miller, E. M. (1994). Paternal provisioning versus mate seeking in human populations.
Personality and Individual Differences, 17, 227-255.
Miller, E. M. (1995). Environmental variability selects for large families only in special
circumstances: another objection to differential K theory. Personality and Individual
Differences, 19, 903-918.
Moon, S. B. (1988). A cross-cultural study of the Kaufman Assessment Battery for Children
with Korean children. Unpublished doctoral dissertation, University of Alabama.
Neisser, U., Boodoo, G., Bouchard, T. J., Jr, Boykin, A. W., Brody, N., Ceci, S. J.,
Halpern, D. F., Loehlin, J. C., Perloff, R., Sternberg, R. J., & Urbina, S. (1996).
Intelligence: knowns and unknowns. American Psychologist, 51, 77-101.
Osborne, R. T. & McGurk, F. (1982). The testing of Negro intelligence. Athens, GA:
Foundation for Human Understanding
Owen, K. (1992). The suitability of Raven's Progressive Matrices for various groups in
South Africa. Personality and Individual Differences, 13, 149-159.
Raven, J. C., Court, J. H., & Raven, J. (1995). Coloured Progressive Matrices. Oxford:
Oxford Psychologists Press.
Rose, S., Kamin, L. J., & Lewontin, R. C. (1984). Not in our genes. New York: Penguin.
Rushton, J. P. (1991). Do r/K strategies underlie human race differences? A reply to
Weizmann et al. Canadian Psychology, 32, 29-42.
Rushton, J. P. (1995). Race, evolution and behavior: A life history perspective. New
Brunswick, NJ: Transaction.
Rushton, J. P., & Ankney, C. D. (1993). The evolutionary selection of human races: A
reply to Miller. Personality and Individual Differences, 15, 677-680.
Scarr, S. (1995). Inheritance, intellgence and achievement. Planning for Higher Educa-
tion, 23, 1-9.
Stevenson, H. W., Stigler, J. W., Lea, S., Lucker, G. W., Kitanawa, S., & Hsu, C. (1985).
Cognitive performance and academic achievement of Japanese, Chinese and
American children. Child Development, 56, 718-734.
Shuey, A. M. (1966). The testing of Negro intelligence. New York: Social Science Press.
Takeuchi, M., & Scott, R. (1992). Cognitive profiles of Japanese and Canadian
kindergarten and first grade children. Journal of Social Psychology, 132, 505-512.
United States Bureau of the Census (1986). Current Population Reports, series P-23, no.
147.
Vernon, P. E. (1982). The abilities and achievements of Orientals in North America. New
York: Academic Press.
Vincent, K. R. (1991). Black/white IQ differences: does age make a difference? Journal
of Clinical Psychology, 47, 266-270.
Waldman, I. D., Weinberg, R. A., & Scarr, S. (1994). Racial-group differences in IQ in
the Minnesota transracial adoption study: a reply to Levin and Lynn. Intelligence, 19,
29-44.
Weinberg, R., Scarr, S., & Waldman, I. (1995). The Minnesota Transracial Adoption
Study; a follow-up of IQ test performance at adolescence. Intelligence, 16, 117-135.
Yerkes, M. (1921). Psychological examining in the US army. Memoirs of the National
Academy of Sciences, 15, 1-890.
Zindi, F. (1994). Differences in psychometric performance. The Psychologist, 7,549-552.
Chapter 14
Intelligence and information processing

/. /. Deary

1. A PERSONAL NOTE

In 1980, for the first time, the University of Edinburgh allowed its medical
students to take a year out to complete an honors degree in psychology, and I
agreed to be a guinea pig. Wishing to complete some background reading over
the summer vacation prior to beginning the intensive year, the first book
recommended by a staff member was Psychology is About People by Hans
Eysenck (1972). Two things about the writer were obvious. He had a knack of
getting one to turn pages; his style was gripping. The writer was a scientist and
wished psychology to be judged by the criteria of natural science. One more
thing: he understood me. I had been sent to the Scottish equivalent of a
grammar school after performing well on the Scottish Qualifying Exam (the
equivalent of the English eleven plus). This notion of having been "spotted" by
a set of reasoning tests, the content of which had nothing to do with school
work, intrigued me, and it is to these tests that I owe my opportunities for
education which would otherwise have been available only at a prohibitive cost.
Therefore, Eysenck's essay on The Rise of the Mediocrity encapsulated my
untutored feelings about the destruction of the U.K.'s grammar school system.
The topic I chose for my research project in psychology was "inspection time
and intelligence," and the first background paper to which I was referred was
Eysenck's 1967 paper in the Journal of Educational Psychology on experimental
approaches to the study of human intelligence. Later, the empirical work from
my undergraduate project appeared as part of a chapter in the book A Model
for Intelligence (Eysenck, 1982), my first published paper.
A year or so later I sought a way to continue the study of human intelligence
during the "elective" period of my medical degree. My psychology project
supervisor—Chris Brand—suggested going to work with Hans Eysenck and
kindly wrote to him by way of introduction. As ever, Chris's introduction was
quirky, informing Professor Eysenck that I wrote modern poetry and tended to
wear black nail varnish and an anarchist badge. However, despite Hans's
replying that he did not understand modern poetry, he arranged for me to
 Intelligence and information processing 283

spend 16 happy weeks at the University of Bath trying to test some of the more
speculative biochemical hypotheses of Elaine and Alan Hendrickson's (1980,
1982) model of synaptic function and human intelligence. I recall my first
interview with Hans clearly. Despite my being a mere medical student, he
offered unlimited time and courteous attention. When we went for coffee in
the Institute of Psychiatry tearoom he bounded up the stairs faster than I could
follow, and he ate a packet of Maltesers with his drink. He seemed to let me
take charge of what turned out to be a long conversation; I hadn't expected
such a fluent and witty writer to be so introverted. He didn't waste words; he
made good use of them.
When I returned to the Maudsley Hospital a few years later as a junior
psychiatrist, I was able to discuss physiological approaches to intelligence with
Hans's then lieutenant Paul Barrett and, when a year later I made a career
switch to academic psychology, it was to intelligence and personality that I
turned for research inspiration. Not only that, but my approach was governed
by an Eysenckian way of looking at things.

2. A WAY OF LOOKING

Some people inspire because they discover something big—like Watson and
Crick and the structure of DNA. Some people inspire because they conceive a
grand theoretical scheme—like, say, Piaget. However, some people just seem
to make the right initial decisions concerning the way to go about things. Such
is Eysenck. First, his approach to the classification of phenomena was
appealing. The use of latent trait analyses to examine the structure of
personality and mental abilities seemed a way of getting at nature's hidden
structures underlying phenotypic variation. What was more impressive was
that, unlike some researchers in intelligence and personality, Eysenck always
saw the classification problem as a first step only. The more essential move was
to discover the psychobiological bases of the traits identified. It may have been
my medical training, but the emphasis on reductionism and the methods of
natural science applied to human variation made the Eysenckian approach
seem like the right one.

3. LANDMARKS

What is striking in reviewing Eysenck's contributions to the study of

information processing and intelligence is his consistency over the years.
Much of what he began to articulate about the construct, measurement, and
analysis of intelligence in 1953 he has stuck to and has found that it requires
repeating and relearning by psychologists decade after decade. For example, it
284 Intelligence

is astonishing how often one needs to reexplain that there is no contradiction

between g (general intelligence), group factors, and primary mental abilities.
Eysenck (1939) was saying this from the first and articulated it clearly in his
popular books (e.g., Eysenck, 1953). However, a misunderstanding, wilful or
otherwise, of the hierarchy of mental abilities forms the basis of much of the
anti-IQ tirade of Gould (1984). And, even the major works by Gustafsson
(1984) and Carroll (1993), reemphasizing the hierarchical structure of mental
abilities, have not permeated to all psychologists, who still prefer to portray
differentialists as a group of warring factions.
Even in his early articles on intelligence, Eysenck (1953) was an advocate of
the scientific method. Taxonomy was all very well, but it could not get at the
mechanisms of intelligence. Like Spearman (1923) had done before him,
Eysenck (1953) began to delve into what might be the components of
intellectual activity. Thus,
Next we might look at the mental operations required to carry out a given task. In one
test we might be asked to learn the content of a passage, in another to remember
something, in yet a third to make inductive judgments, while in a fourth our main task
might be of a perceptual nature. This type of classification is more difficult a priori
because we have little knowledge about the mental processes involved in any
particular mental act, but as a hypothesis we may perhaps let it stand, (p. 29)
From these early, nonempirical guesses as to how to go about unraveling the
nature of individual differences in intelligence emerged a growing devotion to
the nature of mental speed which, Eysenck (1953) correctly adjudged, had
been rather dormant after early bursts of interest at the beginning of the
century (Deary, 1986, 1994) A lasting influence on Eysenck was the work of
Furneaux (1952) who attempted to describe subjects' performances on IQ-type
test items in terms of equations with terms for subjects' speed of solution,
persistence on items, and errors. Furneaux claimed to have found constants in
the value of such relationships as the slope between the log of solution time
and the difficulty level of a problem. He used mental test items themselves as
the analysand and distilled from them some parameters that reflected
individual difference estimates of some important brain functions. Furneaux's
general approach is echoed in Steinberg's (1977) studies of analogical
reasoning items and in the work of Carpenter, Just, and Shell (1990) on
Raven's matrices. However, his studies are, in fact, rarely cited and there is no
train of research leading directly out of his ideas, perhaps because of the
emphasis on mathematical modeling in his papers and because a full treatment
never appeared in a refereed journal.
Perhaps partly as a result of Furneaux's stimulating angle on human ability,
Eysenck (1953) championed two causes related to human ability. First, he
emphasized the influence of noncognitive factors in performance. Persistence,
he believed, was an influence on success, and he described this as a "function
 Intelligence and information processing 285

of personality organization and emotional integration." It is interesting also to

note that persistence (of motives) was the first noncognitive characterological
factor to be identified in Spearman's laboratory at University College London
(Deary, 1996; Webb, 1915) As far as this chapter goes, though, the principal
interest is Eysenck's prescient championing of a mental speed approach to the
investigation of intelligence differences,
on the purely cognitive side, speed of mental functioning emerges as the prime
determinant of intellectual ability. It may be identified with good reason with "g" or
general mental ability or intelligence, (p. 37)
However, at this time there was little contemporary research to go on, and
Eysenck's (1953) formulations were largely intuitive,
When a problem is first perceived there starts in the brain an orderly sequence of
events which results in the production of a chain of "trial solutions." These trial
solutions do not necessarily become conscious, each of them being simply a particular
mode of organization of some part of the brain structure. It is the rate at which these
"modes of organization" are set up, broken down, and remodeled which underlies
the concept of mental speed, (p. 37)
Eysenck has a way of summing up the state of an area and pointing the
forward direction. As his next landmark contribution to information processing
and intelligence I chose his 1967 paper in the British Journal of Educational
Psychology, the first academic article I read (in 1979). First, he summarized
progress in intelligence research: a first phase of "g" led by Spearman and
Binet; a second phase of definition and theory led by Thurstone and
Spearman; a third phase of factor analysis and the identification of group
factors of intelligence with contributions by Thurstone, Thompson, Burt, and
Holzinger; and a fourth phase involving Guilford's model of operations,
contents and products. He was not impressed by Guilford's failure to note the
fact that his tests had almost universally positive correlations,
Guilford has truly cut out the Dane from his production of Hamlet. If this is really the
best model (1965 style) which psychology can offer of intelligence and intellect, then
the time seems to have come to retrace our steps; something has gone very wrong
indeed! (p. 82)
His diagnosis was that,
I would suggest that the psychometric approach has become almost completely
divorced from both psychological theory and experiment, (p. 83)
Furneaux's system was again praised not least because it "reinstates the mental
speed factor to its theoretical pre-eminence as the main cognitive determinant
of mental test solving ability" (p. 84). Furneaux's finding of the log latency
versus item difficulty slope association was hailed as "a constant, one of the few
which exist in psychology." It was the general notion that work such as
286 Intelligence

Furneaux's could reorient the study of intelligence that excited Eysenck; he

suggested that researchers should study speed of processing, notwithstanding
the early, abortive attempts with reaction times. But it is precisely reaction
times that rear their heads in this article and provide the most influential part
of it. Eysenck discussed the work of Merkel (1885), Hick (1952), and Hyman
(1953) which had shown stable individual differences in the slope of reaction
time plots when response times are examined as a function of stimulus
uncertainty. Roth's (1964) study had suggested that people who score more
highly on IQ-type tests had shallower slopes. Of course, this marked the
beginning of much subsequent work on the Hick reaction time procedure and
intelligence (Jensen, 1987a). Sparking off individual lines of research such as
this was only a limited aim of his paper whose main purpose was,
to suggest the importance of starting out on a fifth stage of intelligence assessment, a
new stage based on theoretical and experimental work, and not divorced from the
main body of academic psychology. ... Investigations should pay more attention to
laboratory studies of ... speed of information processing, (p. 96)
This conclusion was similar to Spearman's in 1904 when he commented that
the decision of Binet and Henri to use tests of higher level function would be of
more practical than theoretical value in the study of intelligence. Akin to many
researchers at present, Spearman's student Abelson (1911) attempted to
discover the psychological principles underlying mental test scores. This
complaint was offered to the Binet-Simon test of intelligence,
They do not know what these tests measure or signify. The tests are isolated from the
main body of scientific psychology. They neither derive much light from it, nor do
they import much to it.
This almost identical cry predates Eysenck's by almost 50 years! 1967,
though, was quite a year. Not only did Eysenck publish what Jensen (1986, p.
93) called his "theoretical manifesto" for intelligence, he also published his
Biological Basis of Personality. Interestingly, it was also the year that first saw
Neisser's (1967) epoch-making Cognitive Psychology, a book which provided a
toolkit for those wishing to explore intelligence and information processing
associations.
By 1982, with the publication of A Model for Intelligence, where we see
Eysenck again advocating the joining of psychology's two disciplines
(correlational/differential and experimental), only the "first few, faltering
steps" of the experimental approach to intelligence were being taken (p. 1).
Apart from a historical review of speed approaches to intelligence, the book's
edited chapters offered information about research on inspection times,
reaction times, electroencepalography, biochemistry, components of logical
thinking, and so forth. The book was a phenomenon. It did not sell hugely, or
even well, but it made a large impact on the field of intelligence research. In
the Social Sciences Citation Index alone, Jensen's (1982) chapter on reaction
 Intelligence and information processing 287

times has, to date, over 200 citations, Brand and Deary's (1982) inspection time
chapter over 100, Eysenck's pieces over 100, and the Hendricksons' (1982)
chapters over 100 (when combined). Thus, partly because of Eysenck, by the
early 1980s intelligence and information processing had been conceptualized
as an identifiable approach to human intelligence and some of the key research
now seen by a wide audience. A Model for Intelligence arguably marks the true
beginning of the experimental study of intelligence that Galton and Spearman
saw glimmer, fade, and extinguish.

4. THE CHANGING FACE OF INTELLIGENCE RESEARCH

Arguably, it was said above, but the world of intelligence research was already
changing in the direction of information processing models. Years before,
when the promise of multivariate analyses were first being realized, Sargent
(1942) had warned that "factor analysis is not a substitute for the experimental
analysis of the individual" and had advocated a combined differential-
experimental approach to human abilities. Hunt's series of experiments on the
nature of verbal ability, which combined experimental and differential
approaches, began in the mid-1970s (e.g., Hunt, Lunneborg, & Lewis, 1975)
In 1976, Resnick's edited volume The Nature of Intelligence contained several
chapters that suggested that cognitive-experimental approaches were needed
to fully understand human intelligence differences. In 1977 the journal
Intelligence sprang into life, with Detterman (1977) arguing that there was a
need for a focus for research into intelligence. From the first issue it was clear
that a substantial proportion of the contributors to Intelligence would construe
intelligence from an information processing point of view. This may be seen as
a marriage de convenance between the apparently unprogressive, static
description of abilities provided by the psychometricians and the promise of
cognitive processes being laid bare by the shiny, happy cognitivists. An earlier
proposed attachment between psychometric intelligence and learning theory
had not been consummated (Bachelder & Denny, 1977; Estes, 1974).
One new line of attack was the study of standard psychometric tasks from an
information processing point of view in an effort better to understand the
demands they were making of cognitive processes. Royer's (1977) study of the
Block Design test is one such. This type of approach was carried out with great
ingenuity by Sternberg (1977) in his studies of reasoning tasks and their basic
components.
A broader, emerging approach was to ask whether aspects of constructs
identified by human experimental psychology might have parameters whose
values were associated with individual differences in mental abilities.
288 Intelligence

Intelligence's second volume has two good examples, which asked whether
perception and attention, respectively, might be associated with intelligence.
Royer (1978) introduced his article as follows:
The purpose of this paper is to discuss what the role of perceptual processes in
intelligence is.... If we measure the current status of the question by the indexing of
literature ... it seems that the question does not exist, (p. 11)
Among other issues, Royer examined how the altering of the information
processing characteristics of tasks altered performance on them (c.f. the Hick
task). In anticipation of inspection time research, and despite his principal
interest being stimulus complexity, Royer hypothesized that,
experimentation on the limits of the perceptual system's encoding processes can
provide information about other limits of the information processing system, (p. 33)
Reflecting similar comments made by others he insisted that,
there is a need to investigate intelligence tasks using experimental techniques rather
than correlational ones. I would add that the need is imperative, (p. 37)
It is interesting that Royer (1978) admitted that a critique of his approach
was that he emphasized parallel processing. In the work since then the
overwhelming emphasis has been on serial processing. Like a growing number
of papers that reflected an awakening to the possibly fecund association of
differential and cognitive psychology, Royer's paper was a manifesto—a
statement of encouragement and intent—rather than an empirical contri-
bution. It was a call to arms rather than a battle fought. A similarly theoreti-
cally suggestive but empirically empty paper was contributed by Zeaman
(1978) who urged that the concept of attention could provide a base for
intelligence research.
The general methodological approach I will take to the problem of relating individual
differences in attention and intelligence can be described as an experimental process
analysis, and may be contrasted with the traditional psychometric approach to the
problem.
A quite different approach to the problem makes use of a theoretically guided
experimental analysis of tasks in which attention is presumed to play a role. Controls
for the intrusion of processes other than attention are arranged experimentally. The
requirements of such an approach are first of all a theory in which attention is a
major theoretical construct, and secondly a diverse set of experimental operations to
anchor empirically the inferences of differential behavioural processes with their
accompanying individual differences, (pp. 56-57)
Later in the 1978 volume of Intelligence there was a series of articles on the
possibilities for unifying psychology's two disciplines hi the study of
intelligence. The papers had been presented at the American Educational
Research Association in 1978 in Toronto, Canada. Snow (1978) provided an
 Intelligence and information processing 289

informative series of descriptions of the main psychometric factors in

intelligence and a survey of some possible cognitive approaches. Perhaps the
best article of all was that by Sternberg (1978) who captured and documented
the new movement very well,
There seems to be widespread concurrence among theoreticians and methodologists
alike that new approaches to studying intelligence should somehow combine the
differential and cognitive (information-processing) approaches that have been used
in the past, and that the combination should somehow enable the investigator to
isolate components of intelligence that are elementary (at some level of analysis),
(p. 196)
That parenthetical comment, as we shall see later, was well-advised.
Sternberg's (1978) piece is full of useful information, tantalizing suggestions,
and wisdom, such as the realization that there are many false leads in this area,
which are then not replicated. The history of the area, of course, shows that the
opposite is common too; there are many positive leads that are not followed up
(Deary, 1986, 1994). Sternberg outlined some promising approaches to the
future study of intelligence such as structural equation modeling, cognitive
components of intelligence, cognitive correlates of intelligence, and computer
theories of intelligence. Indeed, perhaps little more than promise could be
expected at this stage for, as Carroll (1978) stated in the same issue,
the state of the art in individual differences research in an information-processing
mode can be thought of as little more than embryonic, (p. 114)
Most of the prophets of the new, combined approach to intelligence had
reservations as well as enthusiasms about the enterprise. An example of some
measured criticism of the approach may be found in the symposium paper by
Hunt and MacLeod (1978). They foresaw some problems for cognitivists who
had tried to apply psychometric techniques and for differential psychologists
who "have tried to interpret their measures in terms of theories of cognitive
psychology" (p. 129). They suspected that there are "deep conceptual
differences between the differential and cognitive psychology approaches to
thought." Though they make it clear they could have used several tasks, they
place their discussion in the setting of the sentence-verification task. For
example, they provide a thoughtful discussion on whether average perfor-
mance on a cognitive task should be considered as a primitive (for example,
overall performance on the sentence verification task, though one could
equally well insert overall performance on the Hick reaction time task or the
Sternberg memory scanning task) or whether one should employ parameters
derived from within the task. The latter, they note, involves the application of a
model to the task performance, which one is then theoretically committed to.
Also, whereas the differential approach tends to use linear models of perfor-
mance "information-processing theories of cognitive processes regard
290 Intelligence

performance as a nonlinear function of primitive variables of the model"

(p. 139). Moreover, they evince concern about the qualitatively different ways
in which subjects approach cognitive tasks.
This worry remains relevant, but not entirely in the way that was articulated
by Hunt and MacLeod (1978). We should ask ourselves whether, if we are
troubled by the qualitatively different ways that subjects approach tasks, we
have truly made contact with information processing primitives. This raises the
issue of the level which researchers have in mind when they are thinking about
information processing parameters. If it is at the level of overall cognitive task
performance then there is a prima facie case for being very skeptical about
whether there is anything primitive about the measure at all. Take the case of
the sentence verification task (Clark & Chase, 1972). There are lawful changes
in reaction time with experimental manipulations of the task, but the task does
not offer an obvious decomposition in terms of information processing
primitives. Indeed, it appears to be complex, with many possible cognitive
processes being involved in its performance. What, if any, isomorphisms exist
between task performance and brain processes remains obscure. At essence,
this problem centers on the validity of the experimental task being used to
study the bases of individual differences in psychometric test performance. If
we respond to Cronbach's (1957) call for a wedding of psychology's two
disciplines, and if we accept the strictures of the cognitivety-oriented
psychologist to the effect that we do not truly understand the nature of
intelligence differences, then the burden of parameterization of the human
cognitive system falls upon the cognitive psychologist. Therefore, for any task
or set of tasks or cognitive theories that we appeal to, we must be firm in asking
to what extent they provide a valid model for cognitive performance. We shall
see below that such valid task decomposition is hard to come by.
In the same symposium-based series of contributions to Intelligence that we
have been discussing, Carroll (1978) also makes this point,
Surveying and carefully examining this literature has caused me to conclude that little
progress has been made thus far in understanding mental abilities in terms of
processes. It can be argued, to be sure, that there has been some success in
identifying psychological processes, but the interpretation of these processes often
stands or falls depending on whether one can accept the information-processing
models on which the identification of a particular process is based. Further, the
experimental identification of a process often depends chiefly upon the finding of
individual differences in the parameters of the process, which has led, in effect, to the
identification of a whole, new series of individual "traits" that are little related to the
mental abilities isolated in classical psychometric studies. Even if the relations are
found to be of substantial magnitude, it is not very revealing or informative merely to
establish the correspondences between traits and processes that are defined largely
on the basis of those traits. There is an obvious circularity in all this. (p. 88)
 Intelligence and information processing 291

Perhaps as a way out of such circularity, we shall see that there are very
different levels of task that are used to search for information processing
primitives. Whereas some have looked to experimental/cognitive paradigms
such as the sentence verification and memory scanning tasks, others have
looked to psychophysical procedures and physiological parameters. The latter
may well have something to be said for it, at least in avoiding the problem of
subjects' qualitatively different strategies in approaching cognitive tasks. For, if
one can identify a basic process which acts as a limiting factor in cognitive test
performance, one may argue that it provides a necessary bottleneck through
which all cognitive processing must pass. Vickers and Smith (1986) describe
this philosophy clearly,
one major strategy guiding attempts to measure the speed of mental functioning has
been to isolate some process sufficiently elementary to be relatively immune from
influence by higher cognitive activities or by motivational and social factors. In its
focus on a simple, component process, likely to play a limiting role in most (if not all)
more complex processes, this strategy resembles the employment of standard
algorithms as benchmark tests of the processing speed of a digital computer, (p. 619)
Let us end this section with a clear statement to the effect that, around the
turn of the decades from the late 1970s to the early 1980s, something new was
happening in intelligence research; something that had been proposed since its
inception, but that had lain dormant for over 60 years. The feel of this time is
captured nicely by Carroll and Maxwell's piece in the Annual Review of
Psychology for 1979, and the tone of the review, written for the general
psychologist rather than the expert audience who read Carroll's Intelligence
article the year before, is upbeat.
A discernible new trend, however, is a budding but fitful and hesitant courtship
between two traditionally separate disciplines of psychology—psychometricians, on
the one hand, and experimental cognitive psychology, on the other. ... this current
trend represents a coming to full circle of tendencies that were evident already
around the turn of the century when J. McK. Cattell, Binet, Spearman and others
attempted, with little real success, to measure intelligence through observations of
simple processes such as sensory discrimination, choice reaction time, and memory
span. (p. 604)
The fresh wind blowing is that of cognitive psychology and the prospect that its
perspective may be able to reform psychometricians and the theory of IDs [individual
differences] in a radical way.... a Phoenix-like revival of directions that were evident
80 years ago. (pp. 633-634)

5. OTHER INFLUENCES

Before picking up the story of the information processing approach to

intelligence and assessing how it has fared since the partly Eysenck-driven kick
292 Intelligence

start in 1982, it is proper to acknowledge that contributions to a field of

research can be made other than by writing influential articles. Apart from this
obvious contribution, there at least two ways in which Eysenck has importantly
influenced the area of research.
First, he has literally and metaphorically provided a forum for the discussion
of information processing and intelligence. His founding (Eysenck, 1980) of
the International Society for the Study of Individual Differences (ISSID) and
its associated journal Personality and Individual Differences (PAID) provided
homes for a homeless topic. Of course, the founding of the journal Intelligence
was equally if not more important. However, before these two journals, and
before ISSID, there was no natural home for the study of information
processing and intelligence. Without such a home an area of study will find it
hard to grow, debate and resolve issues. A good example of the importance of
a forum is the whole issue of PAID in 1986 that was given over to research on
inspection time and intelligence. Ten years after the initial discovery of the link
between inspection time and intelligence (Nettelbeck & Lally, 1976), this issue
provided a substantial foundation for the research that emerged in the 10 and
now 20 years since (Deary & Stough, 1996)
Second, partly by his personal advocacy, partly by his providing a journal
forum, and partly because, unlike most of us he reads German, he has brought
to researchers' attentions the ideas of people who might otherwise be largely
unknown. The important result of Roth (1964) arguably started the interest in
intelligence and the Hick reaction time paradigm. The work of Lehrl (e.g.,
Lehrl & Fischer, 1988) and what Eysenck has called the Erlangen school have
been championed by Eysenck. This school provided a confidently-stated, wide-
ranging, and radically reductionist account of human information processing
that was intended to restructure our thinking about intelligent behavior. The
theory and methodology of this school have not caught on widely, perhaps
because of their premature mathematical parameterization of human
information processing, and partly because of their acceptance of debatable
results of others. Similarly, the wide-ranging theories of people like Weiss
(1989) and Alan and Elaine Hendrickson (1980, 1982) owe some of then-
exposure to Eysenck's editorial efforts and encouragement. All of these writers
were interested in what it is to be intelligent, and have approached the problem
in information processing terms. Their ideas have often been too eclectic for
specialists in any one area to criticize, and some of the expositions have been
downright abstruse and prematurely and suspiciously "complete," but Eysenck
(1980) has always stuck to his view, offered in the editorial of the very first issue
of Personality and Individual Differences, that "we believe in letting the author
tell his own story, without having referees acting as nannies and telling him
how he could tell his own story better."
 Intelligence and information processing 293

6. PRESENT-DAY RESEARCH

It is impossible to track all of the strains of research on information processing

and intelligence from their beginnings in the 1970s/1980s to the present day.
There are too many individual studies and the themes are too numerous.
However, it is first quite possible to declare that the health of the research is
good; the recent contents pages of Intelligence and Personality and Individual
Differences, especially, are witnesses to the continuing attempts to explain
individual differences in cognitive terms. What follows is a survey of the
success of some different approaches to understanding intelligence using
information processing methods. Largely, I have chosen those that Eysenck
has at some time advocated and/or has recently contributed to. However,
because it is distinctly different, I have also included some remarks on
Sternberg's study of cognitive components.

6.1 EEG and evoked potential correlates of intelligence

This is a large research area, which demands considerable technical knowledge
from the reader, and which is published in diverse journals. It was the subject
of an extensive review by Deary and Caryl (1993), who reviewed the entire
literature in this area published up to 1992. They concluded that various EEG
and EP (evoked potential) measures did correlate significantly with psycho-
metric ability test scores, and that significant correlations had been found in
studies of normal adults, not just in studies involving mentally handicapped
subjects and children. However, the factors leading to such associations were
not well understood and, therefore, despite a number of significant
associations, little about the brain/cognitive processes underlying IQ test score
variance had been revealed by such research. Whereas studies of ongoing EEG
had been relatively successful in obtaining relationships with intelligence,
standard EP component latency studies had done less well, though EP studies
emphasizing waveform complexity and component amplitudes had been more
successful.
Deary and Caryl (1993) found one particularly promising strain of research.
Several independent studies had discovered that EP differences associated
with early stimulus processing were related to mental ability test scores
(Blinkhorn & Hendrickson, 1982; D. G. Gilbert, Johnson, B. O. Gilbert, &
McColloch, 1991; Haier, Robinson, Braden, & Williams, 1984; Rhodes,
Dustman, & Beck, 1969; Stough, Nettlebeck, & Cooper, 1990; Zhang, Caryl, &
Deary, 1989). Thus, the poststimulus epoch between 140 and 200 ms appeared
to be related to individual differences in intelligence. In addition, Deary and
Caryl (1993) argued that an atheoretical, descriptive approach was warranted
as the way forward in this research. Much of the theorizing in this field has
been unhelpful and esoteric and an approach that detailed the temporal and
294 Intelligence

topographic associations between brain electrical potentials and particular

mental abilities was to be preferred to one which attempted to make great
theoretical leaps from intelligence test performance to detailed brain
mechanisms. Indeed, the above-noted empirical regularity emerged despite
the very different theoretical orientations and interpretations of the authors.
This regularity has also shown promise in uniting the study of intelligence,
evoked potentials, and inspection time (Caryl, 1994; Colet, Piera, & Pueyo,
1993; Morris & Alcorn, 1995).

6.2 The componential approach of R. J. Sternberg

R. J. Sternberg's approach to understanding the psychological bases of
intelligence test performance was dubbed by him the "cognitive components"
approach to intelligence (Sternberg, 1985). Sternberg (1977) attempted to
discover which components underlie reasoning items involving analogies, and
how they interrelate. He did this, not by correlating reasoning performance
with performances on other information processing tests, but by taking apart
the analogy items themselves.
Sternberg (1977) tested various models to account for subjects'
performances on analogical reasoning test items. He assumed that the
successful completion of such analogy items required the following compo-
nents of performance: encoding of the first and second analogy terms; inferring
a relationship between these items; encoding a third term and then mapping
relations between the first and third terms; and applying the inferred relation
to the third term in order to choose the correct answer option. Kline (1991)
criticized Sternberg's approach, stating that these so-called components are
necessarily correct, i.e. analogy items could not be solved without encoding,
inference, mapping, and application. According to Kline, this made the
components correct a priori and not a matter for empirical demonstration.
However, Kline appeared to have missed the point of Sternberg's original
investigation, because Sternberg did not claim to have demonstrated the
existence of the components empirically. In fact, Sternberg devised a method
which purported to separate the overall item completion times into the times
taken by each component and, further, he attempted to discover whether each
component was applied exhaustively to the analogy items or whether
processing done by each component was halted when a likely solution was
found.
Sternberg (1977) devised a method for presenting parts of each analogy in a
tachistoscope, prior to presenting the whole analogy. This allowed him to
estimate, using simultaneous equations and a "subtraction model" of analogy
response times, the time taken by each component. For instance, if a subject
was shown only the first term of the analogy, e.g. "Lincoln," he or she would be
able to encode this item. Sternberg would allow unlimited time for this
 Intelligence and information processing 295

processing to take place. The whole analogy would then be shown and the
subject asked to respond with the correct answer as quickly as possible.
Sternberg reasoned that the response time for the whole analogy in this case
would be faster than the response time to an analogy originally presented in
complete form, and that the difference between the two response times would
be the time taken to encode the first term. If a subject was shown the first two
terms of an analogy problem, say "LincolnrWashington," he or she would be
able to encode both of these items and then infer a relationship between them,
before seeing the whole analogy and responding with the correct answer.
Seeing the first three terms of the analogy, e.g. "Lincoln:Washington::5:?"
before the entire question was presented would remove the time required for
three encoding operations and the inference and mapping operations from the
response time to the whole analogy. Thus did Sternberg attempt to find out the
proportion of response times taken by the different "components" of
analogical reasoning. (The correct answer to the above analogy is "1," i.e.
the corresponding dollar value of the note that contains the named president's
portrait.)
After testing four different models, Sternberg concluded that the application
and mapping components were self-terminating and that inference might be
exhaustive. However, the difference between the fits of the different models
was often very small, implying that there was little to choose between them.
Sternberg (1977) showed that over 50% of the solution time in verbal analogies
was taken up by encoding of the terms, and that the response component took
a substantial proportion of time. Individuals preselected for their high
reasoning ability appeared to fit the models better and had faster responding
times, but were slower at encoding. The multiple correlation between
reasoning scores and component latencies was over 0.7.
Mullholland, Pellegrino, and Glaser (1980) replicated some of these findings
and demonstrated regularities in the increases in response times as additional
stimulus elements and transformations were added. Perhaps the largest single
body of research on Sternberg's componential method was reported by
Sternberg and Gardner (1983), where the method was applied to analogy,
series completion, and classification items each in verbal, picture, and
geometric forms. This study showed that the components used in Sternberg's
(1977) original study of analogy items were not reliably identified as significant
parameters from different tasks, and that some components emerged from
some tasks but not others. The components of inference, mapping, and
application, whose characteristics appeared to be the main discovery of this
method (Sternberg, 1977), could not be separated reliably in the study by
Sternberg and Gardner, and they had to be combined into a "reasoning"
component. When component latencies were averaged across task content, in
order to compare the three types of task, the mean intercorrelation among
components with the same name in different types of task was 0.32, whereas
296 Intelligence

the mean correlation among component latencies for components with

different names was 0.24. Similar results were obtained when the component
latencies were collapsed across task types in order to compare different task
contents. Sternberg and Gardner argued that this afforded some convergent
and discriminant validity for their componential theory, though Whitely (1980)
had found little evidence for cross-task correlation of corresponding
components.
The poor intercomponent correlations of Sternberg and Gardner (1983)
notwithstanding, there were other considerations that limited the conclusions
that might be derived from this approach. They analyzed the results of four
components: encoding, reasoning, justification, and comparison. First,
encoding often had a poor fit to their data. Second, the reasoning component
was a composite that had locked within it most of the components that were
earlier reckoned to be of interest. Third, the justification component did not
exist in some tasks. Fourth, the comparison component was another
composite, not found in the original list. Alderton, Goldman, and Pellegrino
(1985) found that, even given the correct processing of analogy items, high
ability individuals tended to be helped to the correct answer by perusing the
answer options, whereas lower ability individuals were often distracted away
from the correct answer to an incorrect alternative. In this study, same name
processes across tasks tended to correlate at about the same level as different
level processes.
In summary, the knowledge obtained about the bases of intelligence
differences from the Sternberg subtraction approach to the dissection of IQ-
type test items has not lived up to its early promise. Rarely were competing
models of component function tested competitively to estimate whether one
was significantly better than another. So-called independent processes
correlated significantly in speed as well as accuracy, making a g-type
explanation of the Sternberg results possible. Same-label processes correlated
at only 0.3 across tasks, and sometimes had zero correlation. Although the
chopping up of tasks appeared intuitively correct at first, as Kline (1991)
recognized, this dissection appeared more arbitrary as follow-up studies failed
to demonstrate the existence of the said components as significant parameters
in the models of item solution. It was not clear how generalizable Steinberg's
components were intended to be. As stated by Hunt (1980), the interest in the
components underlying IQ test items lies in the possibility that such compo-
nents will also be important in real world tasks; it was not clear that Steinberg's
components had any life outside the IQ-type items themselves or, indeed, how
the independent existence of such processing components was to be
demonstrated. Whitely (1980) emphasized that Steinberg's approach was not
successful at modeling response accuracy to analogy test items and that, "... no
explicit mechanism is postulated to relate subject differences in the compo-
nents to subject differences in total response time."
 Intelligence and information processing 297

6.3 Hick reaction time, Sternberg memory scanning, and cognitive ability
If the attempt to split IQ test items into basic processing components has not
met to date with conspicuous success, what of the attempts that begin with
components from theoretically well-understood cognitive tasks and attempts to
correlate these to scores on psychometric ability tests? For as long as
psychology has been an empirical endeavor, some form of reaction time has
attracted the attention of those who fancied that they could understand
intelligence in terms of speed of information processing (Galton, 1883;
Wissler, 1901). As early as 1933 Beck had reviewed over 30 studies which had
examined the relationship between reaction time and tests of mental abilities.
In all instances, negative correlations should be taken to indicate that brighter
subjects had faster reaction times. Briefly, Beck found: 14 intelligence
correlations with simple and discriminative reaction times ranging from 0.32
to —0.90 with a median of -0.16 (the high result was the surprising report by
Peak and Boring in 1926 after testing 5 senior students); 14 correlations of
serial reaction time and intelligence with a range of 0.03 to —0.53 and a median
of —0.18; 5 correlations of intelligence and speed of reading giving a range of
-0.14 to -0.32, median -0.30; 6 correlations looking at intelligence and speed
in serial verbal tasks with a range of 0.06 to -0.23, median -0.12; and 6
correlations of intelligence with speed of reflex response latency ranging from
0.08 to —0.24, median —0.06. The review by Beck was incomplete. It did not
include the Travis and Hunter (1928) report of a correlation of -0.87 between
intelligence as measured by the Otis and Iowa exams and the time between
tapping the patellar tendon and the arrival of the motor nerve impulse at the
quadriceps femoris muscle!
In recent years the focus of intelligence researchers has been on reaction
time paradigms with an arguably better theoretical underpinning. Sternberg's
(1966) rapid memory scanning reaction time procedure has attracted some
research attention in the field of intelligence (Chiang & Atkinson, 1976; Deary,
Langan, Graham, Hepburn, & Frier, 1992; Jensen, 1987b; Puckett & Kausler,
1984; Todman & Gibb, 1985) but, although psychometric test scores tended to
correlate significantly with reaction times from this paradigm, the theoretical
difficulties with the Sternberg test, especially with the reliability and
interpretation of its slope, have led to little theoretical advancement about
the bases of intelligence in terms of information processing mechanisms.
Correlations between cognitive ability test scores and Sternberg rapid memory
scanning test parameters have tended to be with overall reaction times or with
the intercept (Todman & Gibb, 1985), but not with the hypothetical
components which the slope was originally thought to comprise (Deary et
al., 1992; Jensen, 1987b).
The reaction time (RT) procedure which has attracted most attention in
intelligence research has been the so-called Hick paradigm. Jensen's (1987a)
review of research in intelligence using the Hick paradigm examined 33 study
298 Intelligence

samples from 27 studies, involving a total of 2317 subjects. Twenty-one of

the samples involved Jensen as author or co-author, approximately half of the
samples included college or university students, and others included above-
average ability children. The average fit of the RT data to Hick's law was 0.995
for decision times, while movement times tended not to fit an increasing slope.
Decision time (DT) will be used here to refer to the time taken to lift the finger
from the home button on the reaction time apparatus after the stimulus light
has been lit (this is called reaction time by Jensen, which is confusing, since the
overall response time has the same acronym), whereas movement time (MT)
will be used to refer to the time taken to press the response button
corresponding to the appropriate stimulus light, after the finger has been lifted
from the "home" button. In the Jensen review, the TV-weighted mean correl-
ations across studies between parameters derived from the Hick paradigm and
tests of mental ability were as follows (with correlations corrected for
unreliability and restricted range given in parentheses): -0.31 (—0.32) for
overall DT mean; -0.18 (-0.25) for DT intercept; -0.18 (-0.28) for DT slope;
-0.32 (-0.48) for DT variability; -0.29 (-0.30) for MT mean; and -0.02
(—0.02) for MT variability. Negative signs preceding the correlations indicate
that brighter subjects were faster and less variable on the Hick RT task
parameters.
Therefore, there were generally modest correlations in the expected
direction with various Hick RT procedure parameters. DT slope, the
parameter originally thought to tap individual differences basic to psycho-
metric intelligence, had no exclusive correlation with ability scores, and the
correlations with slope were not the highest. Even MT was related to
differences in cognitive ability. Jensen (1987b) also argued that there should be
higher correlations between estimates of intelligence and the DTs associated
with greater degrees of stimulus uncertainty. He offered a summary of
evidence from 15 independent groups to show that the correlations between
cognitive ability estimates and DTs for 0, 1, 2 and 3 bits of information were
-0.19, -0.21, -0.24, and -0.26, respectively. Therefore, the differences
between these correlations were very small, as was the absolute size of the
correlations, but they ran in the expected direction, given the hypothesis that
the higher ability person had a greater rate of gain of information. However,
several technically adequate studies did not find this relationship.
Jensen (1987b) concluded that correlations between Hick parameters and
IQ scores were based upon some general speed and/or efficiency reflected in
most aspects of performance in the Hick paradigm. Jensen even considered the
hypothesis that the general factor extracted from a battery of RT tasks might
be the same as the g factor extracted from a battery of psychometric ability
tests. In addition, Jensen attempted to refute explanations of the Hick RT-IQ
correlations based upon the following hypotheses: that there was a common
 Intelligence and information processing 299

test-taking factor; that high ability subjects operated a speed-accuracy trade-

off; and that high ability subjects were characterized by their high motivation or
arousal (see chapter 11).
A series of problems associated with the Jensen apparatus used to estimate
Hick RT parameters was raised by Longstreth (1984) who suggested that there
were order effects, visual attention effects, and response biases associated with
the apparatus. Longstreth also demonstrated that the DT slope-IQ
correlation, and the association between DT complexity and IQ did not hold
when groups of normal subjects were considered. Jensen and Vernon (1986)
and Jensen (1987a) answered much of the criticism raised by Longstreth.
However, perhaps the most worrying criticism by Longstreth was that, in the
typical Hick RT procedure, the trials with the small degrees of stimulus
uncertainty occur first, with the trials associated with the greater degrees of
stimulus uncertainty following in increasing order. Therefore, it might be the
case that the higher IQ individuals learn the RT task faster and therefore, the
correlation between Hick RT slope and IQ might come about because the
learning effect is confounded with stimulus uncertainty.
Widaman and Carlson (1989) administered the Hick RT procedure in the
standard ascending manner and added conditions where the degrees of
stimulus uncertainty were met in descending and random orders. As predicted
by the learning effect hypothesis, the slope for RTs was steepest in the
descending condition, intermediate in the random condition, and flattest in the
ascending condition. More importantly, Widaman and Carlson found that the
correlations with IQ tended to go in the direction that allied IQ with those who
become faster with practice; i.e. IQ appeared to be related more to the rate of
RT increase with practice than with RT components per se. Therefore,
intelligence appeared to be related to individual differences in the rate of
automatization of a new task. Widaman and Carlson suggested that
correlations between Hick RT parameters and cognitive ability might be
ephemeral.
Studies on the Hick reaction time task and psychometric intelligence
continue to be published. However, correlations with single parameters tend to
be relatively low and the correlation with the slope parameter is not especially
high when compared with other parameters. Such results were found by
Beauducel and Brocke (1993) who suggested that Hick's law was irrelevant to
the association between reaction time and psychometric intelligence. Beh,
Roberts, and Pritchard-Levy (1994) extended the Hick paradigm by retaining
up to eight targets in the task but increasing the stimulus uncertainty by having
subjects make up to four responses on any one trial. Again, reaction time and
variability indices correlated significantly with psychometric intelligence.
However, the paper is notable for its lack of discussion about what even this
extension to Hick's law contributes to our understanding of intelligence
differences.
300 Intelligence

In summary, the state of research with the Hick RT paradigm is not

dissimilar to that found with the Sternberg componential approach. There are
some modest, significant correlations between cognitive ability test scores and
Hick RT parameters, but they are not sufficiently tied to theoretically
important aspects to offer much hope that the Hick paradigm will lead to
enlightenment concerning the sources of individual differences in intelligence.
Eysenck himself (Barrett, Eysenck, & Lucking, 1986) found that three key
aspects of the Hick technique were problematic: not everyone's reaction time
data fitted Hick's law; the slope-IQ correlation was not particularly strong
when compared with other, supposedly less theoretically interesting,
parameters; and there was no significant increase in reaction time-IQ
correlations as the stimulus uncertainty increased. Additionally, there remains
the possibility that Hick RT-IQ correlations are due to unexpected aspects of
Hick RT performance, such as rate of improvement on the task. Eysenck
extended the Hick task using the odd-man-out reaction time procedure
(Frearson & Eysenck, 1986; Frearson et al., 1988) In this procedure three
lights are lit on the Jensen RT device; two are adjacent and one is separated by
at least one unlit light from the other two. The target light is this latter odd-
man-out light. RT indices on this procedure have slightly higher correlations
with psychometric intelligence test scores than classical Hick parameters, and
Brody (1992) has suggested that this procedure has promise. However, the
improved correlations appear to have been bought at the price of complicating
the Hick procedure, further removing it from the possibility of theoretical
analysis in terms of basic information processing elements.

6.4 Inspection time

Inspection time (IT; Vickers, Nettlebeck, & Willson, 1972) has been defined
as, "the minimum exposure time needed for observers reliably to identify a
highly evident feature of a stimulus display" (Levy, 1992). The discovery that
there were individual differences in the stimulus duration needed to perceive a
stimulus and make a discriminative judgment about its features to a given level
of accuracy was made by Cattell while he was carrying out investigations for his
Ph.D in Leipzig (Deary, 1986). Cattell had his subjects discriminate various
colors from a standard gray, and he found that there were reliable individual
differences across people and across colors in the times required in order to
make these discriminations (correlations for these results were computed by
Deary, 1986). The possible relationship between perception time and more
general mental ability was only hinted at in the experiments that Cattell
performed on his subjects (Deary, 1986). Griffing (1895-1896), testing groups
of schoolchildren, exposed black letters on a white board using a tachistoscope
to give accurate 100 ms exposures and discovered that the amount of
information extracted from a single brief exposure of a stimulus was related to
 Intelligence and information processing 301

mental ability as estimated by the teachers. The Spot Pattern Test as used by
Burt (1909-1910) in his first empirical study involved the subjects being given a
series of 25 ms exposures of dot patterns arranged in a 5 x 5 matrix. The
subjects were given sufficient exposures until they reproduced the pattern
exactly. The outcome variable was the accumulated exposure time to correct
discrimination, and this correlated at above 0.7 with teachers' estimates of
intelligence. Livson and Krech (1956) tested 22 college sophomores for their
ability to reproduce dot patterns from brief tachistoscopic exposures and found
a correlation of .54 between this ability to scores on the Wechsler Vocabulary
Scale. None of these findings was followed up.
The modern idea of an inspection time (IT) has a history in the idea of the
perceptual moment, that is, the notion that perception operated in a quantal
fashion such that a stimulus must be present from the beginning of a
perceptual sampling period for a sufficient duration in order to be
discriminable. Vickers et al. (1972) developed the accumulator model of
perception to develop the IT index as,
the time required by a S to make a single observation or inspection of the sensory
input on which a discrimination of relative magnitude is based.
Thus, Vickers and colleagues (Vickers et al., 1972; Vickers & Smith, 1986)
suggested that the amount of time required by a subject in order to make a
given discrimination of relative magnitude under given conditions might
represent a stable characteristic of an individual's perceptual performance.
The typical visual IT task involved a subject in discriminating which of two
briefly presented parallel, vertical lines of markedly different lengths was
longer (Vickers et al., 1972). Typically, the lines were presented in a tachisto-
scope and backward masked with a pattern mask to prevent further stimulus
processing. The difference in the lengths of the lines was set to be so large that
the visual angle they subtended was sufficient to make the discrimination easy,
that is, affording perfect performance at longer presentation times. Responses
in IT tasks were typically unspeeded, with accuracy being emphasized over
speed of responding. Therefore, the task was set up to be a relatively pure test
of speed of visual processing, freed from requirements to react quickly or to
make difficult spatial discriminations.
The idea that IT might be a basic limitation to general cognitive
performance, especially among the mentally handicapped, was conceived by
Nettelbeck (Nettelbeck & Lally, 1976). There is little to be gained by discussing
the results of every study which has examined the relation between IT and IQ-
type test scores. There have been a number of qualitative reviews of the
research (Brand and Deary, 1982; Juhel, 1991; Nettelbeck, 1987) and one
meta-analysis of the IT-IQ research (Kranzler and Jensen, 1989). Nettelbeck
(1987) reviewed 16 sets of results, comprising 529 IT estimates obtained from
439 subjects, which had correlated IQ-type scores and IT measures involving
302 Intelligence

visual or auditory stimuli, and found that the average uncorrected correlation
among young nonretarded adults was -0.35. Nettelbeck estimated the
inspection time-IQ correlation at about —0.5 in a sample of the normal
population. He found stronger evidence for a reliable association between IT
and Performance IQ, as opposed to Verbal. However, of the nine studies
which Nettelbeck reviewed in forming this conclusion, only one contained a
sample of normal adults; the others were composed of samples of university
students or mentally handicapped adults. In the meta-analysis by Kranzler and
Jensen (1989), the uncorrected correlations between IT and Verbal and
Performance IQs in adult, nonretarded samples were, respectively, —0.18 and
-0.45. Deary's (1993) study of the association between IT and WAIS-R ability
in 87 adults with diabetes agreed closely with these summary findings. Eysenck
has made a recent contribution to this field of research by demonstrating that
IT might be associated more strongly with g than with fluid intelligence (Bates
& Eysenck, 1993). In his study of 88 people he found a correlation of .62
between IT and overall scores on the Multidimensional Aptitude Battery. This
was principally an empirical rather than a conceptual contribution to the field,
though the short discussion in the article refers to speed-of-processing as the
key construct in IT and RT (decision time).
The view of IT as a simple speed of processing index had not gone
unchallenged. Some visual IT tasks proved prone to cognitive strategies, such
as apparent movement, though it has never been shown that such strategies
cause the IT-IQ association; rather they tend to reduce or eliminate it (Egan,
1994). Two recent surveys of inspection time research have appeared and
indicate that this topic has generated a great deal of diverse interest (Deary,
1996; Deary & Stough, 1996). Rather than attempt to document all of the
detail involved in these accounts, it might be best to summarize their themes.
First, there seems little doubt that the inspection time-psychometric
intelligence association is well established and modest in extent. That is,
there is a strong enough finding to sustain researchers' interest. Most recent
research has been aimed at explaining rather than replicating the association.
Second, the association does not appear to be explained by high-level factors
such as strategy use in the task (Egan, 1994), motivation in the task situation
(Larson, Saccuzzo, & Brown, 1994) or personality factors (Stough et al., 1996).
Third, there is some evidence from an auditory analogue of the IT task that IT
might partly cause later individual differences in intelligence during child
development (Deary, 1995). Fourth, there are evoked potential correlates that
both IT and IQ-type test scores have in common, at between 140 and 200 ms
after stimulus onset (Caryl, 1994), suggesting a common brain processing basis
for individual differences in both IT and IQ-type items. Fifth, the beginnings of
the psychopharmacological influences on IT are being examined
(McCrimmon, Deary, Huntly, MacLeod, & Frier, 1996). However, the fact
that the IT-IQ association is well established and refuses easily to be explained
 Intelligence and information processing 303

away means that the construct validity of IT has attracted much more scrutiny.
Criticisms and empirical problems with the original IT theory have
accumulated (Chaiken, 1993; Deary, Caryl, & Gibson, 1993; Hecker &
Mapperson, 1996; Levy, 1992; White, 1993, 1996). A brief summary of how
things stand at present would be that the theory and rationale for the
inspection time index is undergoing wholesale change. That is, the inspection
time measure, though it may appear straightforwardly to assess the speed of
intake of visual information, is being reexamined to get at the nature of the
process(es) which underlie performance on the task. This is not at all
discouraging; the explication of the construct(s) beneath a measure such as IT
is what must occur for an understanding of the contribution that an
information processing measure can make to intelligence.

7. CONCLUSIONS

The theme of this chapter has been Eysenck's long-standing demand that the
two disciplines of psychology—differential/correlational and experimental/
cognitive—should join together in the study of human intelligence. The story is
a partial success. Eysenck was not alone in his calling for this joining together
and, especially in the late 1970s and early 1980s there was a groundswell of
opinion to this effect. Since then, a number of possible leads have been
pursued, among which the Hick reaction time, Sternberg memory scan,
inspection time, and brain evoked potentials have been mentioned here. To
these, Eysenck (1988) might have added sensory discrimination, marking a
return to the ideas of Spearman (1904). However, the broad endeavor has not
been a total success, and my reasons for saying this are as follows.
First, some interesting leads have ground to a halt. For example, Steinberg's
method of studying the components of intelligence met various problems,
some outlined above, and has not continued with sufficient empirical studies to
answer these. Of course, one can never have any assurance that components
isolated from test items will ever be isomorphic with actual brain processes, but
it is sad to see the effort left unattended. Similarly, Vernon's (e.g., 1987)
battery of theory-based reaction time tasks—some involving short-term
memory constructs—has not been the source of continuing research interest,
though a recent paper has appeared which examines RT tasks in children
(Miller & Vernon, 1996).
Second, differential psychology has taken on tasks that have not proved
amenable to theoretical decomposition in the way that was originally thought.
The Hick reaction time task, the Sternberg memory scan task, and the Posner
letter matching task have not lived up to their promise of containing
unequivocal information processing elements. The expectation that they
might do so may have been naive, but it is none the less disappointing. Studies
304 Intelligence

using these techniques in relation to psychometric intelligence still appear, but

with less theoretical content and an often unstated inability to link the tasks to
cognitive processes. The tasks themselves are of little interest without valid
theory behind them.
Third, to what extent have differential psychology and experimental
psychology really collaborated? Some tasks have been used by differentialists
that clearly originate from the experimental camp, but experimentalists do not
today, nor ever have, devoted much time to the Hick reaction time task. And
there is no clear evidence of a parallel effort on the Sternberg memory scan
task from the two areas of psychology. Inspection time is a partial exception;
those involved in the psychometric study of the measure have kept hi touch
with those carrying out experimental, theoretical, and biological work on the
construct. However, IT has come from the tradition of psychophysics rather
than experimental psychology. In summary, there is little evidence that
cognitive and differential psychologists at the cutting edge of their respective
areas of research are collaborating to pursue individual differences in
intelligence with modern, validated cognitive brain processing measures.
Fourth, there is a danger of some experimental tasks being as theoretically
complex and intractable as the IQ type test items themselves. Some of the
reaction time measures used by Vernon (1987) were essentially timed Yes/No
responses to difficult questions. Work that relates psychometric intelligence
scores to working memory measures merely pushes the problem of explaining
intelligence to another high-level psychological construct indexed by tasks that
look very like some subtests found in intelligence test batteries (Kyllonen &
Christal, 1990).
Fifth, are we keeping up with advances in cognitive psychology, cognitive
science, and neuroscience? Some of the tasks mentioned above have been left
behind as cognitivists have discovered new ways to parse our thinking
processes. For example, at the risk of being called faddish, one should mention
that there are many potential parameters of a neural network that could form
the basis of an attempt to combine cognitive and differential accounts of
intelligence differences (Rabbitt & Maylor, 1991). There is an obvious
dilemma here. We cannot change research programs with every new idea that
emerges from cognitive psychology. However, we should also be wary of con-
tinuing to conduct research with a task/construct that has clearly outlived its
usefulness elsewhere.
Perhaps one is being too pessimistic too soon in building a list like this. This
area of study is young, and there are some robust findings in the area of IT,
RT, and evoked potentials. It should come as no surprise that the explication
and understanding of the associations will take much longer than their
demonstration. In addition, it should not be too surprising to find that the
associations are modest. It is important, though, not to cover up nakedness and
pretend that by borrowing a dusty test from experimental psychology we have
 Intelligence and information processing 305

been gifted with a window on the mind. Getting into the study of information
processing and intelligence means becoming an expert across psychological
disciplines rather than within one, and one should not uncritically accept an
account of any given task. In borrowing concepts, tests, processes, and
constructs from other areas of psychology in order better to understand human
intelligence differences there will be no gift horses: look all of them carefully in
the mouth.

REFERENCES

Abelson, A. R. (1911). The measurement of mental ability of backward children. British

Journal of Psychology, 4, 268-314.
Alderton, D. L., Goldman, S. R., & Pellegrino, J. W. (1985). Individual differences in
process outcomes for verbal analogy and classification solution. Intelligence, 9, 69-85.
Bachelder, B. L., & Denny, M. R. (1977). A theory of intelligence: 1. Span and the
complexity of stimulus control. Intelligence, 1, 127-150.
Barrett, P., Eysenck, H. J., & Lucking, S. (1986). Reaction time and intelligence: a
replicated study. Intelligence, 10, 9-40.
Bates, T. C, & Eysenck, H. J. (1993). Intelligence, inspection time, and decision time.
Intelligence, 17, 523-531.
Beauducel, A., & Brocke, B. (1993). Intelligence and speed of information processing:
further results and questions on Hick's paradigm and beyond. Personality and
Individual Differences, 15, 627-636.
Beck, L. F. (1933). The role of speed in intelligence. Psychological Bulletin, 30,169-178.
Beh, H. C., Roberts, R. D., & Prichard-Levy, A. (1994). The relationship between
intelligence and choice reaction time within the framework of an extended model of
Hick's law: a preliminary report. Personality and Individual Differences, 16, 891-897.
Blinkhorn, S. F., & Hendrickson, D. E. (1982). Averaged evoked responses and
psychometric intelligence. Nature, 195, 596-597.
Brand, C. R., & Deary, I. J. (1982). Intelligence and "inspection time". In H. J. Eysenck
(Ed.), A model for intelligence. Berlin: Springer.
Brody, N. (1992). Intelligence (2nd ed.). New York: Academic Press.
Burt, C. (1909-10). Experimental tests of general intelligence. British Journal of
Psychology, 3, 94-177.
Carpenter, P. A., Just, M. A., & Shell, P. (1990). What one intelligence test measures: a
theoretical account of the processing in the raven Progressive Matrices. Psychological
Review, 97, 404-^31.
Carroll, J. B. (1978). How shall we study individual differences in cognitive abilities?—
Methodological and theoretical problems. Intelligence, 2, 87-115.
Carroll, J. B. (1993). Human cognitive abilities: A survey of factor analytic studies.
Cambridge: Cambridge University Press.
Carroll, J. B., & Maxwell, S. E. (1979). Individual differences in cognitive abilities.
Annual Review of Psychology, 30, 603-640.
Caryl, P. G. (1994). Early event-related potentials correlate with inspection time and
intelligence. Intelligence, 18, 15-46.
306 Intelligence

Chaiken, S. R. (1993). Two models for an inspection time paradigm: processing

distraction and processing speed versus processing speed and asymptotic strength.
Intelligence, 17, 257-283.
Chiang, A., & Atkinson, R. C. (1976). Individual differences and interrelationships
among a select set of cognitive skills. Memory and Cognition, 4, 661-672.
Clark, H. H., & Chase, W. G. (1972). On the process of comparing sentences against
pictures. Cognitive Psychology, 3, 472-517.
Colet, A. V., Piera, P. J. F., & Pueyo, A. A. (1993). Initial stages of information
processing and inspection time: electrophysiological correlates. Personality and
Individual Differences, 14, 733-738.
Cronbach, L. J. (1957). The two disciplines of scientific psychology. American
Psychologist, 12, 671-684.
Deary, I. J. (1986). Inspection time: discover or rediscovery? Personality and Individual
Differences, 7, 625-631.
Deary, I. J. (1993). Inspection time and WAIS-R IQ subtypes: A confirmatory factor
analysis. Intelligence, 17, 223-236.
Deary, I. J. (1994). Sensory discrimination and intelligence: postmortem or resurrec-
tion? American Journal of Psychology, 107, 95-115.
Deary, I. J. (1995). Auditory inspection time and intelligence: what is the direction of
causation? Developmental Psychology, 31, 237-250.
Deary, I. J. (1996). Reductionism and intelligence: the case of inspection tune. Journal
ofBiosocial Science, 28, 405^23.
Deary, I. J. (1996). A (latent) big five personality model in 1915? A reanalysis of Webb's
data. Journal of Personality and Social Psychology, 71, 992-1005.
Deary, I. J., & Caryl, P. G. (1993). Intelligence, EEG and evoked potentials. In P. A.
Vernon (Ed.), Biological approaches to the study of human intelligence. Norwood, NJ:
Ablex
Deary, I. J. Caryl, P. G., & Gibson, G. J. (1993). Nonstationarity and the measurement
of psychophysical response in a visual inspection time task. Perception, 22,1245-1256.
Deary, I. J., Langan, S. J., Graham, K. S., Hepburn, D. A., & Frier, B. M. (1992).
Recurrent severe hypoglycemia, intelligence, and speed of information processing.
Intelligence, 16, 337-359.
Deary, I. J., & Stough, C. (1996). Intelligence and inspection tune: achievement,
prospects and problems. American Psychologist, 51, 599-608.
Detterman, D. K. (1977). Is Intelligence necessary? Intelligence, 1, 1-3.
Egan, V. (1994). Intelligence, inspection time and cognitive strategies. British Journal of
Psychology, 85, 305-316.
Estes, W. K. (1974). Learning theory and intelligence. American Psychologist, 29, 740-
749.
Eysenck, H. J. (1939). Primary mental abilities. British Journal of Educational
Psychology, 9, 270-275.
Eysenck, H. J. (1953). Uses and abuses of psychology. Harmondsworth: Penguin.
Eysenck, H. J. (1967). Intelligence assessment: a theoretical and experimental
approach. British Journal of Educational Psychology, 37, 81-97.
Eysenck, H. J. (1972). Psychology is about people. Harmondsworth, England: Penguin.
Eysenck, H. J. (1980). Editorial. Personality and Individual Differences, 1, 1-2.
Eysenck, H. J (Ed.). (1982). A model for intelligence. Berlin: Springer.
 Intelligence and information processing 307

Eysenck, H. J. (1988). The concept of "intelligence": useful or useless? Intelligence, 12,

1-16.
Frearson, W. M., Barrett, P., & Eysenck, H. J. (1988). Intelligence, reaction time and
the effects of smoking. Personality and Individual Differences, 9, 497-519.
Frearson, W. M., & Eysenck, H. J. (1986). Intelligence, reaction time (RT). and a new
"odd-man-out" RT paradigm. Personality and Individual Differences, 7, 807-817.
Furneaux, W. D. (1952). Some speed, error and difficulty relationships within a problem
solving situation. Nature, 170, 37-38.
Galton, F. (1883). Inquiries into human faculty. London: Dent.
Gilbert, D. G., Johnson S., Gilbert, B. O., & McColloch, M. A. (1991). Event-related
potential correlates of IQ. Personality and Individual Differences, 12, 1183-1184.
Gould, S. J. (1984). The mismeasure of man. Harmondsworth, U.K.: Penguin.
Griffing, H. (1895-6). On the development of visual perception and attention. American
Journal of Psychology, 7, 227-238.
Gustafsson, J. E. (1984). A unifying model for the structure of mental abilities.
Intelligence, 8, 179-203.
Haier, R. J., Robinson, D. L., Braden, W., & Williams, D. (1984). Electrical potentials,
augmenting-reducing and personality differences. Personality and Individual Differ-
ences, 5, 293-301.
Hecker, R., & Mapperson, B. (1996). Sources of individual differences in IT. Personality
and Individual Differences, 21, 697-710.
Hendrickson, A. E. (1982). The biological basis of intelligence Part I: theory. In H. J.
Eysenck (Ed.), A model for intelligence. Berlin: Springer.
Hendrickson, D. E. (1982). The biological basis of intelligence Part II: measurement. In
H. J. Eysenck (Ed.), A model for intelligence. Berlin: Springer.
Hendrickson, D. E. & Hendrickson, A. E. (1980). The biological basis of individual
differences in intelligence. Personality and Individual Differences, 1, 3-33.
Hick, W. (1952). On the rate of gain of information. Quarterly Journal of Experimental
Psychology, 4, 11-26.
Hunt, E. (1980). Intelligence as an information processing concept. British Journal of
Psychology, 71, 449-474.
Hunt, E., Lunneborg, C., & Lewis, J. (1975). What does it mean to be high verbal?
Cognitive Psychology, 7, 194-227.
Hunt, E., & MacLeod, C. M. (1978). The sentence-verification paradigm: a case study
of two conflicting approaches to individual differences. Intelligence, 2, 129-144.
Hyman, R. (1953). Stimulus information as a determinant of reaction time. Journal of
Experimental Psychology, 45, 188-196.
Jensen, A. R. (1982). Reaction time and psychometric g. In H. J. Eysenck (Ed.), A
model for intelligence. Berlin: Springer.
Jensen, A. R. (1986). The theory of intelligence. In S. Modgil, & C. Modgil (Eds.), Hans
Eysenck: consensus and controversy. Philadelphia, PA: Falmer.
Jensen, A. R. (1987a). Individual difference in the Hick paradigm. In P. A. Vernon
(Ed.), Speed of information processing and intelligence. Norwood, NJ: Ablex.
Jensen, A. R. (1987b). Process differences and individual differences in some cognitive
tasks. Intelligence, 11, 107-136.
Jensen, A. R., & Vernon, P. A. (1986). Jensen's reaction time studies: a reply to
Longstreth. Intelligence, 10, 153-179.
308 Intelligence

Juhel, J. (1991). Relationships between psychometric intelligence and information-

processing speed. Cahiers de Psychologic Cognitive, 11, 73-105.
Kline, P. (1991). Steinberg's components: noncontingent concepts. Personality and
Individual Differences, 12, 873-876.
Kranzler, J. H., & Jensen, A. R. (1989). Inspection time and intelligence: a meta-
analysis. Intelligence, 13, 329-347.
Kyllonen, P. C., & Christal, R. E. (1990). Reasoning ability is (little more than) working-
memory capacity?! Intelligence, 14, 389-433.
Larson, G. E., Saccuzzo, D. P., & Brown, J. (1994). Motivation: cause or confound in
information processing/intelligence correlations? Acta Psychologica, 85, 25-37.
Lehrl, S., & Fischer, B. (1988). The basic parameters of human information processing:
their role in the determination of intelligence. Personality and Individual Differences,
9, 883-896.
Levy, P. (1992). Inspection time and its relation to intelligence: issues of measurement
and meaning. Personality and Individual Differences, 13, 987-1002.
Livson, N., & Krech, D. (1956). Dynamic systems, perceptual differentiation and
intelligence. Journal of Personality, 25, 46-58.
Longstreth, L. E. (1984). Jensen's reaction time investigations of intelligence: a critique.
Intelligence, 8, 139-160.
McCrimmon, R. J., Deary, I. J., Huntly, B. J. P., MacLeod, K. J., & Frier, B. M. (1996).
Visual information processing during controlled hypoglycaemia in humans. Brain,
119, 1277-1287.
Merkel, J. (1885). Die zeitlichen Verhaltnisse der Willenstatigkeit. Philosophische
Studien, 2, 73-127.
Miller, L. T., & Vernon, P. A. (1996). Intelligence, reaction time, and working memory
in 4- to 6-year old children. Intelligence, 22, 155-190.
Morris, G. L., & Alcorn, M. B. (1995). Raven's Progressive Matrices and inspection
time: P200 slope correlates. Personality and Individual Differences, 18, 81-87.
Mullholland, T. M., Pellegrino, J. W., & Glaser, R. (1980). Components of geometric
analogy solution. Cognitive Psychology, 12, 252-284.
Neisser, U. (1967). Cognitive psychology. New York: Appleton Century Crofts.
Nettelbeck, T. (1987). Inspection time and intelligence. In P. A Vernon (Ed.), Speed of
information processing and intelligence. Norwood, NJ: Ablex.
Nettelbeck, T., & Lally, M. (1976). Inspection time and measured intelligence. British
Journal of Psychology, 67, 17-22.
Peak, H., & Boring, E. G. (1926). The factor of speed in intelligence. Journal of
Experimental Psychology, 9, 71-94.
Puckett, J. M., & Kausler, D. H. (1984). Individual differences and models of memory
span: a role for memory search rate? Journal of Experimental Psychology: Learning,
Memory and Cognition, 10, 72-82.
Rabbitt, P. M. A., & Maylor, E. A. (1991). Investigating models of human performance.
British Journal of Psychology, 82, 259-290.
Resnick, L. B (Ed.). (1976). The nature of intelligence. Hillsdale, NJ: Erlbaum.
Rhodes, L., Dustman, R., & Beck, E. (1969). The visual evoked response: a comparison
of bright and dull children. Electroencephalography and Clinical Neurophysiology, 27,
364-372.
 Intelligence and information processing 309

Roth, E. (1964). Die Geschwindigkeit der Verarbeitung von Information und ihr
Zusammenhang mit Intelligenz. Zeitschrift fur Experimentelle und Angewandte
Psychologic, 11, 616-622.
Royer, F. L. (1977). Information processing in the block design task. Intelligence, 1, 32-
50.
Royer, F. L. (1978). Intelligence and the processing of stimulus structure. Intelligence, 2,

Sargent, W. (1942). How shall we study individual differences? Psychological Review, 49,
170-182.
Snow, R. E. (1978). Theory and method for research on aptitude processes. Intelligence,
2, 225-278.
Spearman, C. (1904). "General intelligence," objectively determined and measured.
American Journal of Psychology, 15, 201-293.
Spearman, C. (1923). The nature of "Intelligence' and the Principles of Cognition.
London: MacMillan.
Sternberg, S. (1966). High speed scanning in human memory. Science, 153, 652-654.
Sternberg, R. J. (1977). Component processes in analogical reasoning. Psychological
Review, 84, 353-378.
Sternberg, R. J. (1978). Intelligence research at the interface between differential and
cognitive psychology. Intelligence, 2, 195-222.
Sternberg, R. J. (1985). Beyond IQ: a triarchic theory of human intelligence. Cambridge:
Cambridge University Press.
Sternberg, R. J., & Gardner, M. K. (1983). Unities in inductive reasoning. Journal of
Experimental Psychology: General, 112, 80-116.
Stough, C., Brebner, J., Nettelbeck, T., Cooper, C. J., Bates, T., & Mangan, G. L.
(1996). The relationship between intelligence, personality and inspection time. British
Journal of Psychology, 81, 255-268.
Stough, C., Nettelbeck, T., & Cooper, C. J. (1990). Evoked brain potentials, string
length and intelligence. Personality and Individual Differences, 11, 401-406.
Todman, J., & Gibb, C. M. (1985). High speed memory scanning in retarded and non-
retarded adolescents. British Journal of Psychology, 76, 49-57.
Travis, E. L., & Hunter, T. A. (1928). The relation between "intelligence" and reflex
conduction rate. Journal of Experimental Psychology, 11, 342-354.
Vernon, P. A. (1987). New developments in reaction time research. In P. A. Vernon
(Ed.), Speed of information processing and intelligence. Norwood, NJ: Ablex.
Vickers, D., Nettelbeck, T., & Willson, R. J. (1972). Perceptual indices of performance:
the measurement of "inspection time" and "noise" in the visual system. Perception, 1,
263-295.
Vickers, D., & Smith, P. L. (1986). The rational for the inspection time index.
Personality and Individual Differences, 7, 609-624.
Webb, E. (1915). Character and intelligence. British Journal of Psychology Monographs,
13, 1-99.
Weiss, V. (1989). From short-term memory capacity toward the EEG resonance code.
Personality and Individual Differences, 10, 501-508.
White, M. (1993). The inspection time rationale fails to demonstrate that inspection
time is a measure of the speed of post-sensory processing. Personality and Individual
Differences, 15, 185-198.
310 Intelligence

White, M. (1996). Interpreting inspection time as a measure of the speed of sensory

processing. Personality and Individual Differences, 20, 351-363.
Whitely, S. E. (1980). Latent trait models in the study of intelligence. Intelligence, 4,97-
132.
Widaman, K. F., & Carlson, J. S. (1989). Procedural effects on performance on the Hick
paradigm. Intelligence, 13, 63-86.
Wissler, C. (1901). The correlation of mental and physical tests. Psychological Review,
Monograph No. 3.
Zeaman, D. (1978). Some relations of general intelligence and selective attention.
Intelligence, 2, 55-73.
Zhang, Y., Caryl, P. G., & Deary, I. J. (1989). Evoked potentials, inspection time and
intelligence. Personality and Individual Differences, 10, 1079-1094.
Chapter 15
Malleability and change in intelligence

N. Brody

1. INTRODUCTION

Hans Eysenck writes about George Washington Carver as follows:

... a black born in Missouri during the American Civil War, and probably the greatest
American biologist of the last century, despite a background which is a catalogue of
appalling misfortunes and deprivations.
His father died before he was born, the ailing son of negro slaves in the deep
South. His mother was abducted when he was a baby. He was brought up in a
poverty-stricken house by whites who were barely literate. He was denied schooling
because of his colour and had to piece together the rudiments of his education while
performing the most menial tasks. He was constantly hungry, was dogged by ill health
and had a severe stammer thought to have been brought on by childhood traumas.
Yet he succeeded in gaining a formal training—a Bachelor of Science degree in
agriculture—and went on to change the eating habits of the South, and to carry out
original research, working in the field of synthetics (one of the first scientists to do
so), creating the science of agricultural chemistry and laying the foundation for the
United States peanut industry. His discoveries and inventions are legion....
Of the tens of thousands of molly-coddled youngsters receiving higher education in
the United States today, with all their advantages, none is likely to achieve a tithe of
what the self-taught George Washington Carver achieved. (Eysenck and Kamin,
1981, p. 170)
I begin my chapter with this quotation from Hans because it exemplifies one of
the most enduring aspects of his genius—what for want of a better term I shall
call his taste. Eysenck always identifies the central problems in the field and
has an uncanny ability to be decades ahead of everyone else in his insights. The
accomplishments of George Washington Carver direct us to a central
phenomena in the study of intelligence—why do individuals differ who are
exposed to what superficially appears to be a constant environment? More-
over, a moments reflection on Carver's accomplishments should convince us
that his intellectual accomplishments could not have been constant over his
life-span. As a child deprived of formal education and reared in surroundings
bereft of intellectual stimulation, he clearly could not have evinced
312 Intelligence

extraordinary intellectual accomplishments early in life. Yet he was somehow

able to acquire an extraordinary range of intellectual skills as an adult. Clearly
his intellectual knowledge and performance changed from early childhood to
adulthood. In this chapter I discuss change, continuity and malleability of
intelligence.

2. UNEQUAL INTELLIGENCE AMONG EQUAL ENVIRONMENTS

The existence of a range of intellectual accomplishments among individuals

exposed to environments that appear to be constant or that do not appear to
encourage high intellectual accomplishment is a ubiquitous phenomenon.
George Washington Carver's accomplishments may be startling in the
incongruity between accomplishment and environmental setting. They are,
however, merely quantitatively extreme—not qualitatively anomalous. Con-
sider three related findings.

2.1 The Abecedarian Project

The Abecedarian Project is an intensive intervention project designed to
increase the intellectual performance of children being reared in extremely
disadvantaged circumstances (Campbell & Ramey, 1994,1995). The designers
of the project used 13 indices of risk to identify children who were likely to
have poor intellectual development. The risk factors included low parental
education, low or zero family income, father absence, and low maternal IQ. In
the academic community in which the project was conducted, the children who
met the selection criterion for inclusion were overwhelmingly African-
American children whose families had a median income of zero. Many were
children reared in single parent homes living on public assistance whose
mothers were teenagers at the time of their birth. Children chosen for
participation in the project were randomly assigned to a control group or to an
experimental intervention starting shortly after birth and continuing for the
entire preschool period. The intervention included an extensive exposure to a
university based day care experience for several hours a day. There are two
incidental findings of the Abecedarian Project that appear to me to be of some
interest (I shall discuss the main findings of the project when I consider the
malleability of intelligence). Although maternal IQ was one of the 13 variables
used for selection, it was not the main nor was it the most heavily weighted
variable used—as a result maternal IQ varied considerably for the project
participants. Maternal IQ values ranged from 42 to 124. These results remind
us that individuals occupying similar social positions in our society will differ
widely in IQ. Variations in social class background, racial background, income,
employment, characteristics of neighborhoods and the like were minimal for
 Malleability and change in intelligence 313

these women—and even though maternal IQ was used (although not heavily
used) as a basis for inclusion in the study, these women had IQs that differed
by over five standard deviations. Maternal IQ was used as a covariate in
analyses of outcomes of the intervention when the children were 12. Maternal
IQ accounted for approximately 25% of the variance in performance of the
children on tests of intelligence and academic achievement. The conclusion
seems obvious—individuals occupying similar, and in this instance deprived,
environments differ widely in their intellectual abilities and these differences
are related to the acquisition of knowledge and to the acquisition of knowledge
of their children. Moreover, the exposure to an intense relatively constant
preschool age intellectual socialization experience did not eradicate individual
differences in intellectual performance for these children that were predictable
from a knowledge of maternal characteristics.

3. INTELLIGENCE OF DEAF PEOPLE

Consider the results of Braden's comprehensive analysis of the intellectual

performance of deaf people. Braden (1994) noted that the deaf population as a
group experience an environment that could reasonably be assumed to depress
intellectual performance. Deaf people have a high incidence of neurological
problems and often have jobs with low occupational status. For the deaf
children of hearing parents, there is a frequent disruption of the ability to fully
communicate with parents as young children. Deaf people often exhibit deficits
in measures of verbal ability. Deaf people perform as well as hearing people on
tests of nonverbal abstract reasoning ability that are considered to be good
measures of general intelligence. Despite what appears to be environmental
deprivations, deaf people are able to acquire basic intellectual skills that are
comparable to hearing individuals.

3.1 Asian school children

Consider the performance of Asian school children on tests of mathematical
knowledge relative to that of American children. Stevenson and Stigler (1992)
reported the results of a number of well-designed large-scale cross-national
comparative studies of mathematical knowledge. Children attending elemen-
tary schools in China and Japan obtain scores on tests of mathematical
knowledge that are well over one standard deviation higher than that of
children attending elementary schools in America, although the IQs of
children in Asian countries are comparable to those of American children.
Japanese elementary schools provide a relatively constant intellectual
environment for children. Japanese elementary school children are exposed
to a standardized curriculum based on whole class teaching—grouping and
314 Intelligence

tracking are not used. In addition, the schools emphasize a cooperative rather
than competitive approach to learning in which all members of the class are
collectively responsible for the educational achievements of each member of
the class. Despite these commitments to an egalitarian and equal educational
exposure, variability in the performance of Japanese children on tests of
mathematical knowledge are comparable to the variability of performance of
American children who are exposed to much more variable educational
experiences with respect to formal exposure to mathematical concepts.
The three examples considered above reinforce Eysenck's discussion of the
accomplishments of George Washington Carver—individuals exposed to a
constant and even an assaultive intellectual environment will differ widely in
the acquisition of intellectual accomplishments.

4. WHAT CHANGES?

In order to discuss change and continuity of intelligence, it is useful to

distinguish among different conceptual aspects of intelligence. Eysenck dist-
inguishes intelligence A, referring to a biological intelligence consisting of
intellectual genotypes and the influence of prenatal and postnatal biological
events that influence cerebral neurophysiology, from intelligence B and
intelligence C. Intelligence B consists of the modification of intelligence A by a
host of environmental and personality characteristics that shape the develop-
ment of intellectual abilities and the acquisition of intellectual achievements.
Intelligence C refers to the objective measurement of some type of intellectual
knowledge or ability that may serve as a phenotype for scientific analysis.
Measures of intelligence C may differ with respect to the level of their satur-
ation with intelligence A variance.
For the purposes of my analysis, I shall consider a somewhat different but
related tripartite division of conceptual intellectual concepts. I shall distinguish
among intellectual genotypes, hypothetical true levels of intelligence, and
scores on tests of intelligence.

4.1 Intellectual genotypes

There are genotypes that influence the development of intelligence. A person's
genotypes are fixed at the moment of conception. Given current technology,
we shall assume that the genes which influence intelligence cannot be
modified. While genotypes are constant, the genes that influence intelligence
are probably not constant over the life-span and are not invariant in different
environments. If this is correct, it is appropriate to define the genotype for
intelligence by using a double subscript for age-related genotypic influences
and for environmentally specific genotypic influences. Consider some
 Malleability and change in intelligence 315

hypothetical examples of environmental influences. In environments that have

high concentrations of lead, there is a relationship between levels of lead in the
blood and performance on tests of intelligence. For example, in the Port Pirie
study of a community with a lead smelter and relatively high levels of lead
exposures, children whose blood concentrations of lead were in the highest
decile had IQs that were approximately one standard deviation lower than
children with blood concentrations of lead that were in the lowest decile
(McMichael et al, 1988). I assume that variations in lead concentrations in the
blood are influenced by environmental exposures and by heritable variations in
lead metabolism. If this is correct, then the genes that influence the
development of intelligence in an environment similar to Port Pirie's in which
there are high concentrations of ambient lead include those that influence lead
metabolism. In another environment that has low or nonexistent lead
concentrations, genes that influence lead metabolism would not influence
intelligence. The example suggests that the genotypes that influence
intellectual development will vary in different environments.
There are probably cultural variations that also influence the genotypes
which influence intelligence. Different teaching methods and ways of
organizing instructional practices may have different impacts on children
with different personality and temperamental characteristics. The variables
that might differentially determine the genotypes that influence intelligence
may be as varied as the degree of tolerance for aggressive behavior in the
classroom to the use of different instructional methods in the classroom.
Just as the genotypes that influence the development of intelligence may
vary in different environments, the genes that influence the development of
intelligence may be different at different ages. Intelligence declines over the
last half of the adult life-span (see Brody, 1992 for a review of changes in
intelligence over the adult life-span). Genes that influence the process of aging
may be relevant to intelligence in the last half of the life-span although they
may be of little relevance to intellectual functioning among young adults and
children. Both longitudinal and cross-sectional studies of twins indicate that
MZ twins exhibit correlations close to the upper bound reliability of tests and
DZ twins exhibit declining correlations over time (McGue, Bouchard, lacono,
& Lykken, 1993; Wilson, 1986). Since intelligence changes over the adult life-
span, exhibiting accelerating declines over the latter half of the adult life-span,
the relatively constant MZ twin correlation implies that changes for adult MZ
twins must occur in tandem. DZ twins by contrast must exhibit declines that
are idiosyncratic with respect to timing and pattern. If the declines in adult IQ
are partially attributable to biological influences on aging, then genes that
control the aging process will be increasingly relevant to the intellectual level of
an individual over the adult life-span. This analysis may contribute to an
explanation of the relatively low DZ twin correlations for IQ obtained in the
Swedish Adult Study of Aging for performance on tests of intellectual ability
316 Intelligence

obtained from older adults (DZ apart = .32; DZ together, = .22). MZ twins
whether reared together or apart exhibited relatively high correlations for
performance on tests of intellectual ability (MZ apart = .78; MZ together =
.80) (Pedersen, Plomin, Nesselroade, & McLean, 1992).
It is also possible to perform a formal test for the continuity of genotypic
influences of IQ at different ages. Correlations between the intelligence test
performance of biological parents and their adopted children with whom they
have little or no postnatal contact are a function of three parameters—the
heritability of adult IQ, the heritability of IQ for children and the genetic
covariance between the genetic influences on IQ for children and for adults.
DeFries, Plomin, and LaBuda (1987; see also Phillips & Fulker, 1989) studied
relationships between children's IQ and adult IQ for children participating in
the Colorado Adoption Project. Their analyses indicated that the genetic
correlation between childhood IQ and adult IQ increased from .42 for IQ at
age 1 to .75 for IQ at age 4. These results provide estimates for the genetic
overlap between genes that influence the development of intellectual
functioning in very young children and their parents. Since the children and
the parents of these children were relatively young, the genetic covariances
between childhood IQ from ages 1 to 4 and adult IQ might be different for
relationships between older children and older adults.

4.2 True levels of intelligence

Assume that every individual has a true level of intellectual competence that
varies over time. The true level of intelligence of a person may be defined as
the error-free score of an individual on the complete panoply of intellectual
tasks that are age appropriate for an individual. The use of age appropriate as
a qualifying term draws attention to the obvious notion that intelligence is
assessed (and must be conceptualized) as something that changes over the life-
span and increases greatly from birth to young adulthood. Although we may
use a common score to define a person's intelligence (IQ), it is obvious that our
methods of measurement change from childhood to adulthood. Unlike height,
intelligence is not assessed using the same instrument at different ages. The
absolute level of a person's intelligence increases dramatically from childhood
to adulthood. Constancy in measures of intelligence does not refer to constant
performance on tests of intelligence—rather it refers to constancy of the age
equivalent percentile rank for performance on tests administered at different
ages.

4.3 Obtained scores

How can we reach conclusions about the true level of intelligence of a person?
The true level of intelligence of a person is a hypothetical construct whose
 Malleability and change in intelligence 317

value may be inferred from an individual's performance on tests of intelligence.

There is a remarkable, and by no means self-evident, empirical observation
first noted by Spearman in 1904 that provides a strong empirical foundation for
inferences about the true level of intelligence of a person from performance on
tests of intelligence. Measures of intelligence generally consist of behavioral
observations of the performance of an individual when confronted with a task
(an individual item on a test) that poses some type of intellectual challenge to
the individual. The types of tasks that may be used to assess the intellectual
functioning of an individual are extraordinarily diverse. Nevertheless, they
exhibit a common property of forming a positive manifold in which an
indefinitely large number of tasks are all positively correlated with each other.
It is a mathematical consequence of the Spearman-Brown prophecy formula
that the aggregate performance on a set of items that form a positive manifold
becomes congruent with the hypothetical construct that represents the
common variance in the set of items. The degree to which the aggregate
performance on the items is congruent with the true common variance of the
items is a function of two parameters—the average correlation among the
items and the number of items. As long as a test of intelligence samples widely
among diverse items that belong to the indefinitely large pool of items that
form a positive manifold in the intellectual domain, the aggregate performance
on the items will be extremely close to the true score common variance. If we
define the true level of a person's intelligence as the true aggregate
performance on all possible age appropriate tasks that measure intelligence,
it is mathematically necessary that the obtained test performance is an accurate
index of the true score. Following Spearman (1904), we can call the
hypothetical true score general intelligence and use the letter g to designate
this construct. Studies of changes in intelligence test performance which must
be accurate measures of g (as long as the items contained in the test form a
positive manifold, and they do) may be construed as studies in changes in true
levels of intelligence.

5. KINDS OF CHANGES

5.1 The overlap model

Changes in intelligence test performance include short-term variability in
performance and variability over more extended periods. Writing in 1953,
Eysenck summarized the results of studies of stability of IQ prior to adulthood
as follows: "for a highly reliable ... test like the Stanford-Binet ... we may
expect a correlation of about .95 between test and retest with only a few days
intervening ... If test and retest are separated by a year the correlation will drop
to .91 ... For every additional year intervening, the correlation will drop by .04
... after ten years it will be down to .55" (Eysenck, 1953, p. 68). Eysenck goes on
318 Intelligence

to embrace Anderson's theory of uncorrelated increments in IQ (Anderson,

1946). If increments in a given year are uncorrelated with the average level of
intelligence attained, test-retest correlations decline as the time between
administrations increases. As the cumulative attainment of intellectual ability
increases over time, the ratio of the size of the uncorrelated increment in
ability to the size of the attained ability level will decrease. This implies that the
prediction of adult intelligence will be increasingly accurate from scores
attained at older ages. The overlap in attained ability for intelligence measured
at two different times increases for measurements obtained with a constant
number of intervening years as the age of a person at the initial measurement
increases. If the correlation between measurements taken at two different
times is a function of the overlap in the ability that is measured, then the test-
retest correlation will increase for measures obtained with a constant number
of intervening years as the age of the individual at the initial measurement
increases.
Eysenck's summary of the empirical relationships obtaining among tests of
intelligence administered at different times in a person's life is as accurate
today as it was when he wrote it in 1953. There are, however, a number of
respects in which contemporary research adds to our understanding of changes
in intelligence. Eysenck's analysis of changes in intelligence embraces a model
of random fluctuations in intelligence. Some of the unpredictability in
attempting to predict adult intelligence from measurements obtained in
childhood can be eliminated by aggregating scores obtained on more than one
occasion. For example, Pinneau's analysis of the results of the Berkeley
Growth Study—a longitudinal investigation of the development of
intelligence—indicated that the average IQ based on tests given at ages 5, 6,
and 7 correlated with an age 17 and 18 aggregate .86. Thus there is
considerable stability of IQ from early childhood to the end of high school. The
correlation between an average of IQs obtained at ages 11, 12, and 13 and the
average of IQs obtained at ages 17 and 18 is .96—suggesting that there is
virtually no change in the true level of intelligence between the beginning and
end of high school when aggregates are obtained that remove error variance
associated with single occasion measurements (Pinneau, 1961).

5.2 Phenotypic changes dictated by genotypes

The Anderson overlap model assumes that changes in intelligence are
unrelated to previous levels of intelligence. While this model may constitute
an adequate approximation of the patterns of test-retest correlations for IQ,
distinctions among conceptually different aspects of intelligence permit a more
comprehensive and conceptually accurate understanding of changes in
intelligence. Longitudinal behavior genetic analyses indicate that the herit-
ability of IQ increases over the adult life-span. There are two kinds of studies
 Malleability and change in intelligence 319

that provide clear evidence for the increasing heritability of IQ phenotypes

from early childhood to young adulthood. The Louisville twin study is a
longitudinal twin study of changes in intelligence. Correlations for IQ test
scores for MZ and DZ twins exhibited increasing divergences from 3 months of
age to age 15. MZ twins exhibited increases in the correlation for IQ that reach
an asymptote at age 4 that are close to the short term test-retest reliability of
IQ. Correlations for DZ twins peaked at age 3 in the high 70s and declined to
the low 50s at age 15 (Wilson, 1986). The increasing divergences in correlations
in the IQs of MZ and DZ twin pairs imply that changes in IQ over this period
are those that increase the relationship between the phenotype for IQ and the
genotype for IQ. Changes in IQ are not random—they are determined in part
by the characteristics of the genotype present at the moment of conception.
Evidence for an increase in the relationship between genotype and
phenotype is also obtained in longitudinal studies of adoption. For example, in
the Texas Adoption Project, correlations between the IQs of adoptive parents
and children declined from an initial testing and a second testing obtained 10
years later; correlations between the IQ of biologically unrelated siblings
reared together declined, and correlations between biological mothers and
their children adopted shortly after birth exhibited small increases over this 10-
year period (Loehlin, Horn, & Willerman, 1989). These findings imply that the
heritability of IQ increased over the 10-year period in this study. If the
heritability of IQ increased, this implies that changes in the IQ phenotype were
those that increased the congruence between phenotypes and genotypes that
influence IQ. The direction of changes that occurred were influenced by
characteristics of persons present at the moment of conception. Thus, the
phenotype for intelligence—an IQ test score which is by virtue of the existence
of a positive manifold and the Spearman-Brown prophecy formula an accurate
index of the true level of intelligence, becomes increasingly congruent with an
age-specific genotypic level of intelligence over the life-span.

5.3 Changes predictable from infant measures

In addition to the evidence of behavior genetics research, there is an additional
set of findings that casts doubt on the validity of the Anderson overlap model.
Test-retest correlations that fit the model include measures obtained in
infancy that are not highly correlated with measurements obtained later in life.
This finding fits the Anderson model. Infant intelligence was traditionally
assessed using clinical measures that were heavily weighted toward sensory-
motor abilities. There are contemporary longitudinal studies of the develop-
ment of intelligence from infancy to early childhood employing laboratory
measures of infant cognitive ability. Several investigators have used measures
of the rate of habituation to a constantly presented stimulus as a measure of
infant cognitive functioning. Longitudinal studies relating infant measures
320 Intelligence

obtained neonatally or as late as age one to childhood IQs obtained at ages

three to eight have reported correlations ranging from .3 to .5. Since the infant
measures have relatively low short-term test-retest reliabilities, the disatte-
nuated correlation between infant measures obtained during the first year of
life and childhood IQ may be close to .6 or .7, suggesting that early childhood
IQ is predictable from performance on measures obtained during the first year
of life. (See Columbo (1993) for a review and theoretical discussion of this
literature; see McCall & Garriger (1993) for the results of a meta-analysis of
the correlations obtained in these studies.) In the only study published to date
in which infant measures are related to the IQs of older children, Rose and
Feldman (1995) found that a measure of habituation obtained prior to age 1,
correlated with scores on the first principal component of a battery of tests of
intellectual ability obtained at age 11, .50 (Rose and Feldman, personal
communication). The study of the predictability of adult IQ from measures of
infant cognitive ability is in its infancy. Studies that aggregate performance on a
battery of infant measures obtained on several occasions have not been
reported. It is possible that aggregate indices of infant performance obtained
during the first year of life might be highly predictive of later intellectual
functioning. The available literature is certainly compatible with the notion
that there are precursors of later intellectual development that determine the
course of intellectual development. These results imply that a model of
increments in intellectual development that are uncorrelated with prior
intellectual performance is probably incorrect. They are, however, compatible
with the kinds of conceptual distinctions fundamental to Eysenck's under-
standing of intelligence. The infant measures are probably saturated with
intelligence A variance (or possibly, in my terms, with genotypic influences on
intelligence) that influences the development of intelligence C (scores on an
IQ test).

6. MALLEABILITY

6.1 Is intelligence fixed?

Scores on tests of intelligence are predictable from knowledge of a person's
genotypic characteristics (think of the results of studies of MZ twins reared
apart or together that imply that a person's score on a test of intelligence may
be predicted from knowledge of the score attained at the same age from his or
her MZ co-twins score). Scores on IQ tests are also predictable (although less
accurately) from knowledge of the IQ test scores of a person's biological
parents whether or not these parents rear the person, and from knowledge of
performance on tests of infant cognitive functioning. These findings suggest
that performance on IQ tests is determined by characteristics of persons that
are already in place by the end of the first year of life. Does this imply that
 Malleability and change in intelligence 321

performance on IQ tests is "fixed" by characteristics present at age one? In one

sense the answer is yes. If high predictability is obtained, this implies that
absent some radical alteration in the environment to which an individual is
exposed, future IQ test performance is determined by characteristics present
prior to conception (parental IQ), at conception (genotypes), and by age one
(the cumulative influences that determine intelligence A and B that have
influenced early intellectual development).
Are there environmental encounters that will alter the course of
predetermined intellectual development? In order to answer this question it
is useful to distinguish between encounters that are part and parcel of the
normal course of development and those that are socially engineered to
deliberately alter the course of intellectual development.

6.2 Naturally occurring changes in intelligence

Moffitt, Caspi, Harkness, and Silva (1993) administered IQ tests to a cohort of
children in New Zealand when they were 7, 9,11, and 13. They found that for
the majority of children, changes in scores on successive administrations of the
IQ test were compatible with a model of random fluctuations attributable to
measurement error. They also found that approximately 10% of the children in
their sample had changes in IQ that were larger than expected given the
unreliability of measurement inherent in any single occasion assessment of
intelligence. They obtained comprehensive information about the family
characteristics and experiences of these children as well as information about
crises, and medical and behavioral problems. They compared the character-
istics of the children on 37 variables that might theoretically influence the
development of intelligence and they found that these variables did not reliably
distinguish children whose trajectories of increase or decrease in test
performance were unusual. While individuals may exhibit declines, increases,
or plateaus in their performance on repeatedly presented tests of intelligence,
these differences are not predictable from knowledge of the environmental
encounters of individuals that might theoretically be expected to influence the
development of intelligence.

6.3 Effects of schooling

There is one environmental event that is known to have a dramatic impact on
the development of intelligence—formal schooling (for a review of the effects
of schooling on intelligence see Ceci, 1991). Children who are deprived of
formal schooling exhibit declines in intelligence. Dramatic evidence for this
assertion derives from a study conducted in Prince Edwards County in
Virginia. The Board of Education of this County chose to close the public
schools from 1959 to 1964 rather than comply with a court order to end a
322 Intelligence

segregated school system in which Black students were forced to attend

separate schools. Relative to students from similar background who attended
schools in a neighboring community, it was estimated that children declined
approximately six points in IQ for each year of education that they missed
(Green, Hofmann, Morse, Hayes, & Morgan, 1964).
The effects of formal education on the development of IQ are also seen in
studies that compare amounts of education for children who differ in age. Most
school systems rely on a formal cut-off birthdate for school entry. As a result,
children who differ by one day of age may differ by one year of the amount of
formal education they receive. For example, Cahan and Cohen (1989)
compared the effects of age variations for children who were in the same
grade to variations in amount of education for children who differed only
slightly in age. They found that variations in formal schooling had a more
profound impact on IQ than age variations—the ratio of the effects of a year of
additional schooling on intelligence to the effects of a year of age on
intelligence was approximately 2.21 to 1 and the effects were larger on
measures of fluid intelligence than on measures of crystallized intelligence.
Thus, exposure to formal education is a more powerful influence on
intellectual development than age-related changes in intellectual development.
We also know that at the extremes, the quality of educational exposure may
influence the development of intelligence. Jensen (1977) studied Black
children attending schools in impoverished areas of the American South
prior to the widespread implementation of desegregated schooling and found
that older siblings had lower IQs than their younger siblings and estimated that
IQ declined by 1.42 points per year for this sample. He attributed this decline
to the educational experiences of these children. Other research among
African-American children attending what were presumably more adequate
schools in California failed to find evidence of a systematic decline in
intelligence attributable to the effects of inadequate schooling (Jensen, 1974).
Variations in the amount of formal education attributable to different
starting times are likely to be of diminishing significance as a person grows
older. Consider two children who differ in age by one day. Child A has
completed the first grade and Child B, the older, has completed the second
grade. The difference in the amount of education they have received is
considerable. When Child A has completed the fifth grade and Child B the
sixth, the relative difference in the amount of formal education they have
received has declined and the expected effect of variations on amount of
education on measures of intellectual competence will also decline. This is
precisely what was found in a study reported by Morrison, Smith, and Dow-
Ehrensberger (1995). They found that older children who were one year older
than younger children differed in performance on tests of reading skills by .9
standard deviation units prior to the start of formal schooling. After one year
of formal education, the older children who had completed the first grade
 Malleability and change in intelligence 323

differed from the younger children by 2.36 standard deviation units. After
completing the second grade, the older children outperformed the younger
children by .36 standard deviation units. These results indicate that variations
in intelligence and in what children learn as a result of exposure to differing
amounts of formal education dissipate with the passage of time. Formal
education is almost always necessary for the development of intellectual
competence. But as individuals encounter somewhat more comparable
amounts (and kinds) of formal education, variations in intellectual compe-
tence attributable to variations in education decline.
The ability of individuals to develop their intellect as a result of exposure to
formal education provides a partial explanation for the declining significance
of the effects of shared family environments as individuals age. Adoption
studies typically find that the IQs of adopted children are related to the IQs of
their adoptive parents and to the IQs of their nonbiologically related siblings in
early childhood. The influences of the shared family environment decline as
children grow older and approach zero for children being reared in
environments that differ widely in terms of the intellectual characteristics of
adoptive parents. It should be noted that these studies exclude families at the
extreme lower ends of the social structure, for example, the homeless, the
addicted, the unemployed, families on public assistance and the like are not
well represented in contemporary behavior genetic studies of adoption. As
children are exposed to the relatively constant environment of public
education, the effects of variations in rearing environments attributable to
intellectual socialization experiences associated with the characteristics of the
family that rears a child diminish. Schools create a relatively constant
environment for the actualization of intellectual ability—families by contrast
differ considerably in the intellectual contribution to variations in the
development of intelligence but their influence fades as the schools become
the principal agency of intellectual socialization. As schools become more
uniform (as they are for elementary education in Asian countries that rely on
highly structured curriculum and whole class teaching), variations in what
children learn and in the development of their intellectual abilities may be
assumed to become increasingly determined by variations in their genotypes
that influence the ability to acquire knowledge from more or less constant
environmental exposures.
Intellectual genotypes influence the growth of intellectual competence and
academic achievements for individuals exposed to formal education.
Behavioral genetic analyses of the covariances between IQ and educational
accomplishments indicate that the covariance between IQ and academic
achievement is substantially mediated by common genetic influences. The
correlation between IQ and academic achievement is higher among MZ twins
than DZ twins and is higher for biologically related siblings reared together
than for biologically unrelated siblings reared together. Most of the formal
324 Intelligence

behavioral genetic analyses obtain estimates of the common genetic

contribution to these covariances that exceed .5. Thus the common influence
on intelligence and academic achievement is primarily a genetic one rather
than exposure to a common family (Wadsworth, 1994).

6.4 Preschool interventions designed to increase intelligence and

academic achievement
Can we change intelligence and its influence on education by formal
interventions? There are two kinds of studies focusing on the preschool
period that provide an answer to this question. Headstart investigations study
the effects of intellectual interventions associated with day care provision for
children being reared in poverty. Several studies of Headstart have included
longitudinal investigations of the effects of random assignment of children to
Headstart or to a control group. Exposure to Headstart interventions that
generally last for two to three years leads to increases in IQ and early
intellectual achievements (Consortium for Longitudinal Studies, 1983). The
effects fade two to three years after the end of the project. There may be long-
lived effects of exposure to Headstart experiences—children in Headstart are
less likely to be placed in special education classes and they are more likely to
graduate high school than children not exposed to Headstart. But, on measures
of intelligence and on measures of academic achievement there are no long-
lasting discernible effects of Headstart.
Headstart may be a less than optimal intervention. Headstart programs
rarely last for more than two or three years. Perhaps more intensive
interventions starting shortly after birth and continuing for the entire
preschool age period would have more long-lived influences on intelligence
and academic achievement. There are two such studies whose results are now
available for analysis—the Milwaukee Project and the Abecedarian Project.
The Milwaukee project randomly assigned 20 African-American children
whose mothers were extremely poor and who had low IQ to an intensive
intervention beginning at age 3 months and continuing for the entire preschool
period. The children were exposed to a university-based day care facility for
several hours a day that had a very favorable teacher to child ratio beginning
with a child-tutor ratio of 1:1 and later increasing to a ratio of 3:1.
Interventions were also provided for mothers whose children were in the
experimental treatment group. After many years, the results of this
investigation have been published and are available for public scrutiny.
Writing in 1981 prior to the availability of the full outcomes of this research,
Eysenck noted that at ages eight to nine, children in the experimental group
had IQs that were 20 points higher than children in the control group—a very
substantial effect (Eysenck & Kamin, 1981). Eysenck noted somewhat
skeptically, "children may have been trained to answer the specific questions
 Malleability and change in intelligence 325

on which they were tested; and in any case IQs cannot be reliably tested at low
ages. Furthermore, and most important, the children have not yet reached
maturity and until their final IQs at the age of 16 or so are known, we cannot
really say very much about this experiment." (Eysenck & Kamin, 1981, pp. 57-
58). Eysenck's skepticism was justified. The results presented by Garber (1988)
indicated that the benefits of the intervention declined with the passage of
time. At the end of the intervention the children in the experimental group
scored 32 points higher on the Stanford-Binet than children in the control
group. At age 14, the children in the experimental group had IQs that were 10
points higher than those in the control group—still a substantial increase
attributable to the intervention. The increase in IQ associated with this
intensive intervention may be partially artifactual and partially attributable to
specific training on items that are similar to those included in tests of
intelligence. Children in the experimental group did not differ from children in
the control group on tests of reading or mathematical achievement. Normally
children who differed in IQ by two-thirds of a standard deviation would also
differ in performance on tests of academic achievement. The children in the
experimental group had math scores at the 10th percentile of the Metropolitan
Achievement Test—a result which is close to the expected value of children
whose IQs were comparable to those in the control group (approximately 80).
While the results of the Milwaukee Project are less than impressive,
justifying Eysenck's initial skepticism, there is a second somewhat analogous
project that provides somewhat more promising data with respect to the
possibility of creating preschool interventions to increase IQ. The results for
the Abecedarian Project for project participants at age 12 and 15 have now
been published (Campbell & Ramey, 1994, 1995). The IQ advantages of the
experimental as opposed to the control group declined from approximately
one standard deviation in early childhood and stabilized at approximately one
third of a standard deviation from age 8 through age 15—although it should be
noted that the difference in IQ between the experimental and the control
group was not statistically significant at age 15. Since the IQ difference was
approximately of the same magnitude as it was at earlier ages, the lack of
significance at age 15 must be attributable to increased heterogeneity of scores
at this age. The one third standard deviation increase in IQ was matched by a
comparable increase in tests of academic achievement. The Abecedarian
Project is the first clear demonstration of a sustained increment in intellectual
performance attributable to preschool age interventions.
The outcomes of the Abecedarian Project are convincing and encouraging.
Yet the results also provide information about the limits of preschool
interventions to remediate the diverse consequences of being reared in
extreme poverty in the United States. Consider some of the additional findings
of the study. Additional interventions involving the random assignment of a
home/school teacher to act as an intermediary between the child and the
326 Intelligence

school had no effect on academic achievement or intelligence. At age 12, the

children in the experimental group were compared to a representative sample
of children of the same age attending schools in the same community who were
not selected as being at risk of academic failure. The children in this
comparison group were somewhat above average in social privilege—many had
parents with academic or professional backgrounds—reflecting the com-
position of children attending schools in an academic community. The children
in the comparison group scored approximately one standard deviation higher
on tests of academic achievement and intelligence than children in the
experimental group. Thus the experimental intervention resulting in a one
third standard deviation increase removes approximately 25% of the difference
between the least privileged children in the community and a representative
comparison group of children. Variations in maternal IQ accounted for
approximately three to four times more variance on measures of academic
achievement and intelligence than the influence of the experimental
intervention. These results suggest that preschool interventions cannot
eliminate differences in academic and intellectual performance associated
with variations in parental and demographic characteristics of children.

7. AN INTERPRETATIVE SUMMARY AND CONCLUSION

7.1 Malleability reconsidered
Intelligence is malleable—within limits. Intense and enduring preschool inter-
ventions appear to increase intelligence by one-third of a standard deviation.
While studies of adoption suggest that shared family influences on intelligence
decline as adoptees age, at the extremes, variations in the conditions of rearing
influence intellectual development. For example, Capron and Duyme (1989)
used a complete cross-fostering design to study the effects of relatively large
variations in social class background on the IQs of 14-year-old French
adoptees. The adoptees reared in homes with foster parents of privileged
backgrounds had IQs that were 11.65 points higher than the adoptees reared in
homes with parents with low social status. Turkheimer (1991) reanalyzed their
data taking account of between- and within-group variations in educational
levels of adoptive parents. His regression analysis indicated that the IQ of
adopted children increased by one point for every two years of increase in the
number of years of education completed by the adoptive parent. This result
suggests that the upper bound range of reaction of children to relatively large
variations in the social class backgrounds of the parents who reared them in
contemporary Western societies is approximately one half of a standard
deviation.
Schooling can also influence intellectual development. Children deprived of
schools exhibit dramatic declines in intelligence. Children who attend extreme-
ly poor schools also exhibit declines in intelligence.
 Malleability and change in intelligence 327

7.2 Intellectual resilience. Intelligence as the creator of environmental

influences
Intelligence is malleable—but it is also resilient. To a considerable extent,
individuals who differ in their genotypic levels of intelligence influence the
environment they encounter. Children who have a heritable ability to acquire
knowledge from formal educational settings are likely to enjoy schooling and
intellectual activities and to persist in school. Rehberg and Rosenthal (1978)
investigated the relationship between IQ and social class background on the
decision to continue education beyond high school. They studied all of the
students who entered the ninth grade of a high school and used a longitudinal
design to investigate changes in their plans to continue education after high
school. Their final analysis dealt with actual enrollments in postsecondary
education—the earlier measurements dealt with educational plans. They found
that IQ was slightly less important than a variety of social class background
factors in determining plans to continue education when the children were in
the 9th grade. Parental social class background, parental expectations for
continued education, and peer college intentions all had marginally higher
correlations with a measure of plans to continue education than the correlation
between IQ and plans to continue. Data on actual enrollments exhibited a
different pattern of relationships. The correlation between IQ and the actual
decision to enroll in higher education was now marginally higher than the
correlations between social class, parental expectations, and peer intentions
and actual enrollment. Individuals with high IQ did better in school, received
more encouragement from counselors, may have taken more rigorous courses,
and encountered changing peer influences. IQ influenced the nature of the
environment they encountered. And the decision to continue education will
also undoubtedly influence the nature of the intellectual environment that
individuals encounter after graduation from high school.
Jencks (1979) reviewed surveys of the effects of social class background and
scores on IQ tests on the amount of education acquired and found that IQ was
more predictive of the number of years of education completed by a person
than his or her social class background. In addition, the influence of IQ was
observed in studies of brothers who were reared in the same family. Brothers
with higher IQ were likely to obtain more formal education than their brothers
with lower IQs.
The number of years of education completed by an individual is related to a
person's occupational status. Since IQ influences the amount of education a
person obtains, it also relates to a person's occupational status and mobility.
Jencks's analyses indicated that brothers who had higher school age IQs not
only obtained more education than their brothers with lower IQs, but they also
had higher occupational status. Just as the influence of IQ on the decision to
continue ones education changes over the course of a person's high school
career, the influence of IQ on a person's occupational status changes over the
328 Intelligence

course of his or her working life. Wilk, Desmarais, and Sackett (1995) studied a
large sample of young adults who were fully employed for a five-year period.
They assigned the job held by the person at the start of their study to one of 10
complexity categories. The complexity categories differed with respect to the
average IQ of individuals assigned to jobs similarly classified with respect to
complexity and to the occupational status of the job classification. They studied
changes in job classifications for a five-year period. They found for each of
their initial categories of job classification, that individuals who increased in
the complexity of the job they held and as a result in the occupational status of
their job tended to have higher IQs than individuals who declined in the
complexity of the job they held. These data support the "gravitation"
hypothesis—individuals gravitate to jobs that are commensurate with their
initial levels of ability.
Jencks's analyses of the role of IQ in education and intergenerational
occupational mobility as well as the detailed analyses of the impact of IQ on
schooling and changes in occupation contained in the Wilk et al. and Rehberg
and Rosenthal studies help us to understand the reciprocal relationship
between IQ and the environment. Individual differences in IQ influence an
individual's encounters with the environment and help to shape and define that
environment. The transactions between IQ influenced environmental
encounters and the IQ phenotype may cumulatively influence the IQ
phenotype and may change it in the direction of increasing its relationship
to the IQ genotype. Intelligence is not defined solely by the social world we
encounter—it defines the intellectual world we create for ourselves. It is this
process of extended interaction between IQ and the social world that enables
us to begin to understand the accomplishments of someone like George
Washington Carver. As Western societies become more egalitarian providing
greater opportunity for individuals to be exposed to adequate schooling,
individual differences in the ability to profit from those exposures will
increasingly determine the intellectual accomplishments of individuals.
Contemporary research on continuity and changes in intelligence informs us
of what Hans Eysenck has always believed—scores on IQ tests are related to
an underlying biological reality that influences the trajectories of our lives.
Research on the continuity, change, and malleability of intelligence reinforces
our appreciation for what is at the heart of Hans Eysenck's profound
contributions to knowledge—it is our biological characteristics that individuate
us and that govern the ways in which we relate to our social worlds. Individual
differences related to genotypic characteristics define and shape our
transactions with the environment. None of us is a passive recipient of the
environment that we initially encounter.
 Malleability and change in intelligence 329

REFERENCES

Anderson, J. E. (1946). The prediction of terminal intelligence from infant and

preschool tests. Yearbook for the National Society for the Study of Education, 39, Part I.
Braden, J. P. (1994). Deafness, deprivation, and IQ. New York: Plenum.
Brody, N. (1992). Intelligence (2nd ed.). San Diego, CA: Academic Press.
Cahan, S., & Cohen, N. (1989). Age versus schooling effects on intelligence
development. Child Development, 60, 1239-1249.
Campbell, F. A., & Ramey, C. T. (1994). Effects of early intervention on intellectual and
academic achievement. Child Development, 65, 684-698.
Campbell, F. A., & Ramey, C. T. (1995). Cognitive and school outcomes for high risk
African- American students at middle adolescence: Positive effects of early
intervention. American Educational Research Journal, 32, 743-772.
Capron, C., & Duyme, M. (1989). Assessment of the effects of socioeconomic status on
IQ in a full cross-fostering design. Nature, 340, 552-553.
Ceci, S. J. (1991). How much does schooling influence general intelligence and its
cognitive components? A reassessment of the evidence. Developmental Psychology,
27, 703- 722.
Columbo, J. (1993). Infant cognition: Predicting later intellectual functioning. Newbury
Park, CA: Sage.
Consortium for Longitudinal Studies (1983). As the twig is bent ... lasting effects of
preschool intervention. Hillsdale, NJ: Erlbaum.
DeFries, J. C., Plomin, R., & LaBuda, M. C. (1987). Genetic stability of cognitive
development from childhood to adulthood. Developmental Psychology, 23, 4-12.
Eysenck, H. J. (1953). Uses and abuses of psychology. Baltimore, MD: Penguin Books.
Eysenck, H. J., & Kamin, L. (1981). The intelligence controversy: Eysenck versus Kamin.
New York: Wiley.
Garber, H. L. (1988). The Milwaukee Project: Preventing mental retardation in children at
risk. Washington, DC: American.
Green, R. L., Hofmann, L. J., Morse, R. J., Hayes, M. E., & Morgan, R. F. (1964). The
educational status of children in a district without public schools. Bureau of
Educational Research, College of Education, Michigan State University.
Jencks, C. (1979). Who gets ahead! The determinants of economic success in America.
New York: Basic Books.
Jensen, A. R. (1974). Cumulative deficit: A testable hypothesis. Developmental
Psychology, 10, 996-1019.
Jensen, A. R. (1977). Cumulative deficit in IQ of blacks in the deep south.
Developmental Psychology, 13, 184-191.
Loehlin, J. C., Horn, J. M., & Willerman, L. (1989). Modeling IQ change: Evidence
from the Texas Adoption Project. Child Development, 60, 993-1004.
McCall, R. B., & Garriger, M. S. (1993). A meta-analysis of infant habituation and
recognition memory performance as predictors of later IQ. Child Development, 64,
57-79.
McGue, M., Bouchard, T. J., Jr., lacono, W. G., & Lykken, D. T. (1993). Behavioral
genetics of cognitive ability: A life-span perspective. In R. Plomin & G. E. McLearn
(Eds.), Nature, nurture, & psychology. Washington, DC: American Psychological
Association.
330 Intelligence

McMichael, A. J., Baghurst, P. A., Wigg, N. R., Vimpani, G. V., Robertson, E. R, &
Roberts, R. J. (1988). Port Pirie cohort study: Environmental exposure to lead and
children's abilities at the age of four years. New England Journal of Medicine, 319,
468-475.
Moffitt, T. E., Caspi, A., Harkness, A. R., & Silva, P. A. (1993). The natural history of
change in intellectual performance: Who changes? How much? Is it meaningful?
Journal of Child Psychology and Psychiatry, 34, 455-506.
Morrison, F. J., Smith, L., & Dow-Ehrensberger (1995). Education and cognitive
development. Developmental Psychology, 31, 789-799.
Pedersen, N. L., Plomin, R., Nesselroade, J. R., & McLearn, G. E. (1992). A
quantitative genetic analysis of cognitive abilities during the second half of the life-
span. Psychological Science, 3, 346-353.
Phillips, K., & Fulker, D. W. (1989). Quantitative genetic analysis of longitudinal trends
in adoption designs with application to IQ in the Colorado Adoption Project.
Behavior Genetics, 19, 621-658.
Pinneau, S. R. (1961). Changes in intelligence quotient: Infancy to maturity. New York:
Houghton-Mifflin.
Rehberg, R. A., & Rosenthal, E. R. (1978). Class and merit in the American high school:
An assessment of the meritocratic and revisionist arguments. New York: Longman.
Rose, S. A., & Feldman, J. (1995). The prediction of IQ and specific cognitive abilities
at 11 years from infancy measures. Developmental Psychology, 31, 685-696.
Spearman, C. (1904). "General Intelligence" objectively determined and measured.
American Journal of Psychology, 15, 201-293.
Stevenson, H. W., & Stigler (1992). The learning gap. New York: Summit Books.
Turkheimer, E. (1991). Individual and group differences in adoption studies of IQ.
Psychological Bulletin, 110, 392-405.
Wadsworth, S. J. (1994). School achievement. In J. C. DeFries, R. Plomin, & D. W.
Fulker (Eds.), Nature, and nurture in middle childhood. Cambridge, MA: Blackwell.
Wilk, S. L., Desmarais, L. B., & Sackett, P. R. (1995). Gravitation to jobs commensurate
with ability: Longitudinal and cross- sectional tests. Journal of Applied Psychology, 88,
79-85
Wilson, R. S. (1986). Continuity and change in cognitive ability profile. Behavior
Genetics, 16, 45-60.
PART III

Further Eysenckian interests

Introduction: Hans Eysenck—A man of many
talents

H. B. Gibson

In this section, nine people have written about Eysenck's involvement with
various aspects of psychology and related subjects, but there is much more to
be described about his many and diverse interests. When writing his biography,
which was published in Gibson (1981), I interviewed people up and down the
country and recorded a great wealth of material about their experiences of
working and interacting with Eysenck in different circumstances. I also
engaged in correspondence with various people in the U.K. and abroad, and
even had an interviewer visit his mother in Paris.
Unfortunately, the people I interviewed were all, on the whole, favorably
disposed, or at least neutral in their attitude to Eysenck; the ones who were
hostile generally refused to be interviewed, and most of them did not answer
my letters. Seeking more material, I asked him if he would nominate some of
those whom he knew to be hostile to him and very critical of his work.
Among others he gave me the name of Dr X, a noted psychoanalyst, and
said, "He'll hate my guts, but it might be worth your while contacting him." I
wrote to Dr X and received the following letter in reply:
Dear Dr Gibson,
Thank you for your letter of February 5th. Have you got clearance from Professor
Eysenck to protect those who may comment on him from any reaction for slander or
libel?
Yours sincerely,
I replied pressing for further details, but got no enlightenment beyond a
remark in a further letter that, "I do not think evaluation of Eysenck's work
can be separated from Eysenck as a person." Hostile critics were reluctant to
make any definite statements.
I did not rely entirely on people's reports on Eysenck but consulted as much
documentary evidence about his career as I could find, even going back to the
records of Pitman's College in London where he had studied as a boy of 18.
Gradually, a fairly comprehensive picture of the man of many talents emerged.
 Introduction: Hans Eysenck—A man of many talents 333

Most of the chapters in this book concern Eysenck's academic work, but
there is another side to him: that of a devoted athlete. In the autobiographical
account that he gives in his contribution to Lindzey's^l History of Psychology in
Autobiography (1970) he writes of his time in the Islington air raid precautions
in the early part of the war, and describes how the London cockneys regarded
him as an enemy alien with considerable distrust and some hostility. They were
eventually won round to a greater tolerance of him not by anything he said on
the political front but by his prowess as a batsman in their cricket team. A
German who was very good at cricket surely couldn't be too bad a chap after
all!
The sport to which Eysenck is seriously wedded, however, is tennis. In the
1970s a Canadian psychologist, David Cox, a tennis player of champion
standard and formerly coach to the men's team of the University of British
Columbia, came to work in Eysenck's department at the Institute of Psychiatry
and was much impressed by his involvement with tennis. How seriously tennis
formed a part of Eysenck's life can best be appreciated by Cox's own statement
(1979) which he was kind enough to write for me:
Eysenck is capable of playing aggressively at the net and does not hesitate to punch
away a set-up. The consequence of this style of play is that he is more effective as a
singles player although he certainly enjoys playing doubles. His attitude on the court
is excellent and he would be described in tennis terms as being "mentally tough." He
rarely gets upset when he makes an error and generally shows little emotion on the
court ...
A final point would be to evaluate the importance of tennis in Eysenck's life. It is
very apparent to me that any account of his work which does not consider this aspect
of his daily existence will have missed out something which he himself considers very
important. I use the term "daily" as, by his own admission, he would play tennis every
day for several hours if he had the chance. If unable to play tennis he would choose to
play squash, and then if that were not possible, badminton. In fact, his ability to move
back and forth between these three racquet sports on different days is quite
remarkable as the elements of one, for example the use of the wrist in badminton,
would be considered detrimental to tennis. In conversation with Eysenck it has
become clear to me just how important it is to him to be able to get some exercise
each day preferably in the form of a tennis match. To a person who does not play
tennis this might be seen in terms of an unimportant whim which might be satisfied if
the weather is sunny and warm. However, as a tennis player I understand his passion
and appreciate how important it is to him. He will play under most conditions, and at
morning coffee each day will evaluate the weather and consult with his colleagues to
make sure that a game has been arranged. This daily game, beginning usually about
12.30, takes precedence over virtually all activities. They are arranged around the
time of tennis rather than vice versa. Also, it seems likely that the time on the tennis
court is used sometimes as a period in which he thinks over projects he is working on,
although he would prefer that this did not happen as his game usually suffers as a
consequence.
334 Further Eysenckian interests

This was written when Eysenck was in his sixties, and it can hardly be
expected that now he is in his eighties he will be such an active athlete.
Nevertheless, he still takes a keen interest in sport as a spectator, and he is an
ardent supporter of the Manchester United football team. On his eightieth
birthday he attended a match played by his team, and his presence among the
spectators was noted. This led to birthday greetings being broadcast over the
public address system, as they were proud to acknowledge the presence of so
famous a supporter.
It might be thought that Eysenck's deep involvement with athletics is a thing
apart from his concern with psychology, but this is not the case. He brings his
concept of personality into his involvement with competitive sport. When
David Cox, whom I have quoted above, originally wrote to him from Canada
mentioning that the young players he was coaching had not done as well in
international tennis as might be expected, Eysenck replied as follows, making
use of his concept of P as a dimension of personality associated with com-
petitive aggressiveness:
I imagine that the trouble with your tennis players in Canada is that their P score is
not high enough. I imagine that a dash of that is needed to make somebody fully
competitive in the modern world. Lack of P has always been my trouble; on one
occasion that I won a National Tournament (at the age of 16) against a much more
highly fancied opponent who was on the verge of the Davis Cup Team, this only
happened because he sported a swastika badge and behaved in a very insulting
manner to me: this really put up my dander and for once I was absolutely determined
to win. (Letter, 1978)
Cox remarked that it was odd that Eysenck described himself as being
normally quite low on competitiveness, and expressed the view that anyone
who had challenged him in any way would regard him as being quite the
opposite! I have heard it reported by those who have played tennis with him
that he held that if you really want to win you should refrain from the perfectly
normal human tendency to congratulate your opponent if he makes a
particularly skillful shot; for according to the theory of operant conditioning
you are thereby strengthening tendencies in him that will bring about your own
defeat.
For Eysenck, life is a battle in all its aspects—academic pursuits, sport,
human relationships, and sexual love. He has been fortunate with his endow-
ment in all these spheres and he has certainly striven hard in them all to make
the best of his many talents. We may see him as "a man of many talents" but I
think that he sees himself in another light: to him these various gifts are but
aspects of a very distinctive whole, and it is only the viewpoint of those who
study his work and activities that create an apparent diversity.
He has given his autobiography the title of Rebel with a Cause (Eysenck,
1990), but having studied it, and indeed reviewed it (Gibson, 1991), I am still at
a loss to see precisely what the cause is. He would probably say, the triumph of
 Introduction: Hans Eysenck—A man of many talents 335

truth over error, but because of his perception of science as a battle-ground,

and his endeavor to be a very efficient warrior, I have the impression that, as he
is human like the rest of us, he persuades himself that the banners he so
valiantly supports are for the "right," and those of the enemy are for the
"wrong." On a number of occasions (and I wish that I could give appropriate
references here) he has said that " ... in science a theory only dies because
some more powerful theory comes along to supplant it. We would like to think
that it is always theories of a greater explanatory power replacing those of
lesser value, but I am afraid that the history of science demonstrates that the
popularity of a theory, even among the scientific community, depends to a
large extent on the Zeitgeist of the prevailing culture. Dreadful examples of
this were the general acceptance of the most absurd pseudo-scientific theories
during the Nazi era in Germany and in the Stalinist regime, even among
intelligent men of science."
The sense that Eysenck is a "rebel" is that he has habitually backed those
individuals who have stood out against the accepted establishment opinion: for
instance, Michel Gauquelin the investigator of cosmic rhythms and
psychophysiology (and also a champion tennis player!), Carl Sargent the
paranormal investigator (Eysenck & Sargent, 1982), Philip Burch whose
investigations of the evidence that smoking causes lung cancer set the cat
among the pigeons in the medical world (Burch, 1976), and Grossarth-Maticek
whose researches into the prevention of cancer and heart attacks have been
repudiated by most medical authorities (Grossarth-Maticek & Eysenck, 1991).
Chapters 5, 22, and 23 of this volume address some of these themes. Eysenck's
combative stance has led him on certain occasions to make some rather
intemperate statements about fellow scientists. Thus when referring to the 186
leading scientists, including 18 Nobel prize winners, who signed a statement
entitled "Objections to Astrology" he described these people as "fanatics."
What he actually wrote was:
What we thus have is not astrologers versus scientific sceptics, but two sets of fanatics,
one believing on the basis of suspect evidence in the truth of astrology, the other on
the basis of complete ignorance of what evidence there might be for disbelieving in
astrology. (Eysenck, 1990)
From this statement we would infer that Eysenck has evidence that each and
every one of these scientists has "complete ignorance" of the issues involved in
assessing the claims of astrologers. I do not think that many people would
regard this as probable. It seems more likely that the statement they signed
touched a deep chord in him, and a man "in love" is liable to make intemp-
erate statements.
The trouble about being "a man of many talents" is that no one man can be
an expert on everything, and it is likely that there will be people of fewer talents
who are nevertheless more expert and better informed about some subjects
336 Further Eysenckian interests

than Eysenck. It seems to me probable that among the aforesaid 186 scientists
who were interested to sign the statement about astrology there would be at
least some who were better informed about the issues involved than Eysenck.
To condemn the lot as "fanatics" having "complete ignorance" of the subject
seems to me unwise and likely to lower scientists' respect for his opinion on
other matters.
Having gained the reputation for being an enfant terrible championing a wide
and disparate variety of anti-establishment causes, has meant that even when
Eysenck is discussing issues on which he has a special degree of competence,
critics are less likely to take him seriously. He likes to have a finger in a large
number of pies, but as he has only the same number of fingers as the rest of us,
the result is unfortunate when too many different sorts of pies are on the menu.
I have argued elsewhere (Gibson, 1991) that Eysenck is by nature a highly
extraverted person who has the extraverted propensity to fall in love all too
readily and too passionately with too many "mistresses." Here I am speaking
figuratively, for I do not mean human mistresses—having made the great
mistake in his early life of marrying an unsuitable partner, when that short-
lived marriage was over, he has been unfailingly faithful to his second wife,
despite all the pretty women at the Maudsley who have set their caps at him.
His affairs of the heart have been the topics mentioned in the three parts of
this book, and many others. But alas, when the papers have been dispatched to
the journals and the MSS of the books to the publishers, by the time of the
proofs are ready for checking, our Don Juan of many talents has become so
absorbed in wooing another "mistress" that he does not observe the caution
contained in the old song:
It's gude to be merry and wise,
It's gude to be honest and true;
It's gude to be off with the old love
Before you are on with the new
Eysenck has confessed to a slap-dash tendency that has had results which
people of a less extraverted tendency would find personally disturbing:
Having finished something—an article, a book, a theory—I want to get on to some-
thing else, without carefully going over things again and again, making painstaking
corrections, reading proofs conscientiously, and generally making sure that every-
thing is ship-shape and Bristol fashion. (Eysenck, 1990)
The result of this failing has been that critics of far less ability than he, have
been able to attack his work by pointing out a number of minor discrepancies
in the published work in order to denigrate the whole.
Being a man of so many talents, Eysenck is unsure of what next will capture
his fancy, and he uses the highly expressive simile of the Catherine wheel: The
Catherine wheel in my mind keeps sparking away and it is difficult to know what it
may set on fire. Having such a wide range of interests also poses a problem for
 Introduction: Hans Eysenck—A man of many talents 337

his memory: having read and written so much in so many different fields, and
had a life so packed with hectic incident, the price he has had to pay is to
sacrifice some clarity and accuracy in his episodic memory. In reviewing his
autobiography I have commented on the inaccuracy of his memory for certain
incidents with which I have been personally concerned, and for which there is
documentary evidence. This is not the result of the "benign forgetfulness" of
later life, nor is it the result of giving himself a positive halo, for sometimes he
does not do himself full justice in his inaccurate version of events. Perhaps part
of the story is also found in his habit of dictating, rather than painstakingly
writing up, his many works (see chapter 24). His strange lapses of memory
were apparent nearly 20 years ago when he sent me the draft of his contri-
bution to A History of Psychology in Autobiography (Lindzey, 1970) and he was
most grateful to me for putting him right on so many points when his memory
had failed him.
Whilst thanking me for receiving a copy of my recent book about memory in
later life, he complained that his memory was getting very bad and said that he
believed it was due to "a shortening of the neurons." I do not think that any such
physiological explanation is necessary, for the longer we live and continue to cram
into our memory-bank a huge assortment of diverse material, as is the case with
him, so it becomes more and more difficult to retrieve items at will from so
complex a neuronal network. To use the simile of the library, if one is constantly
adding books of a diverse nature—books on personality, conditioning, behavior
therapy, astrology, ESP, creativity, psychopathology, intelligence, and goodness
knows what else—the task of painstakingly cataloguing them into categories and
subcategories, and endless cross-referencing becomes immense. The huge task of
librarianship within Eysenck's skull is not made easier by his propensity to leap
from one topic to another before, as he says, "generally making sure that
everything is ship-shape and Bristol fashion." An extraverted tendency?
Here we come to a point of controversy between him and me. I have given it
as my opinion that he is of a highly extraverted nature, and I will not trouble to
repeat my arguments here. He, on the other hand, has always maintained that
he is rather introverted. On some points he appears to be introverted, and if
you see him quietly sipping orange juice in a corner at a party while the others
are noisily socializing as they knock back the hard stuff you may think of Item
14 on the EPQ: "Can you usually let yourself go and enjoy yourself at a lively
party?" I would therefore liken Eysenck to Oscar Wilde, who when his hostess
saw him standing all alone in a corner at a party, went up to him and inquired
solicitously, "Are you enjoying yourself Mr Wilde?" He replied, "Oh yes—
after all who else is there here to enjoy?" Perhaps this often explains Eysenck's
apparent withdrawal from society.
In writing my Introduction to Part III of this book, some may think that I
have been overcritical of Eysenck. But I have left it to the authors of the
various chapters to assess his talents in various different fields. I have not yet
338 Further Eysencldan interests

read their chapters so all I can do is to comment on the consequences of

someone having so many talents. He could not possibly have achieved all that
he has in the space of one lifetime if he had been a slow, plodding, and careful
worker. In Britain we needed our Eysenck; he more than any other man this
century has put psychology on the map here, and we owe him an immense debt
of gratitude.

REFERENCES

Burch, P. R. J. (1976). The biology of cancer: A new approach. Lanchester U.K.: MTP.
Cox, D., personal communication dated 24 December 1979.
Eysenck, H. J. (1970). Autobiographical sketch. In G. Lindzay (Ed.), A history of
psychology in autobiography (Vol. VII). San Francisco, CA: W. H. Freeman.
Eysenck, H. J., Letter to D. Cox, dated 24 January 1978.
Eysenck, H. J. (1990). Rebel with a cause: The autobiography of Hans Eysenck. London:
W. H. Allen.
Eysenck, H. J. & Sargent, C. (1982). Explaining the Unexplained. London: Weidenfeld &
Nicholson.
Gibson, H. B. (1981). Hans Eysenck: The man and his work. London: Peter Owen.
Gibson, H. B. (1991). Review of H. J. Eysenck's autobiography. Personality and
Individual Differences, 12, 1117-1119.
Grossarth-Maticek, R. & Eysenck, H. J. (1991). Creative motivation behaviour therapy
as a prophylactic treatment for cancer and coronary heart disease. Behaviour
Research and Therapy, 29, 1-16.
Lindzey, G. (Ed.) (1970). A history of psychology in autobiography (Vol. VII). San
Francisco, CA: W. H. Freeman.
Chapter 16
Classical conditioning and the role of personality

/. Martin

I. INTRODUCTION

There are many unique aspects to classical conditioning, one being that it can
claim a hundred years of history and research. Within this period and within
this topic there have been numerous shifts in interest. Subsequent to Pavlov's
exploration of a wide range of innovative techniques, behavioral theories of
classical conditioning were developed in the U.S.A., first within a rigorous
behaviorism (Hull, 1943), and later within a cognitively oriented era (Rescorla
& Wagner, 1972). Both have generated a solid structure of research based on
very different approaches to learning, the one emphasizing the growth of
stimulus-response (S-R) bonds ("habits"), the other, the acquisition of knowl-
edge and the nature of the representational processes. At the same time there
has also been remarkable progress in neurobiological research. It has become
clear that vertebrate and nonvertebrate animals can provide powerful model
systems to explore the anatomical, cellular and molecular mechanisms of
learning. The search for the "engram" (or engrams) has vastly increased
knowledge of the brain mechanisms underlying classical conditioning (Thomp-
son et al., 1976).
In other areas, interest has declined during these years. Human classical
conditioning, a popular topic in the 1950s and 1960s, has dropped to such a low
level that the question could be asked whether it was now merely of historical
interest (Hugdahl, 1995). Another diminished role is that of the emotions in
conditioning. Reference is made in passing to unconditioned stimuli (UCSs) as
being "biologically significant" but contemporary theories emphasize the
animal's more "cognitive" capacity to acquire knowledge of stimulus and event
relationships (e.g., Rescorla & Wagner, 1972). There is also little or no refer-
ence to individual differences in contemporary animal or human research. Yet
from the beginning, and even within animal research, their influence has been
recognized. Pavlov observed that some of his animals fell asleep in the
conditioning harness while others remained alert. His explanation of these
differences was framed in terms of what was known then about cortical
excitation and inhibition and "weak" and "strong" nervous systems. Eysenck
used these and other Pavlovian concepts (e.g., transmarginal inhibition), as
340 Further Eysenckian interests

well as Hullian concepts of reactive inhibition and conditioned inhibition, to

formulate a theory that stressed the interaction of extraversion-introversion
(E-I) with experimental stimulus conditions conducive to the build-up of
inhibition. Insofar as Eysenck's theory was based on such concepts it could be
said to have been limited by current knowledge in the behavioral and
physiological domains, but he has always been ready to accommodate new
findings into his thinking. Thus, in 1967 he recast his theory to take account of
new data on the reticular formation, terms such as cortical arousal now
replacing the earlier Pavlovian ideas on cortical inhibition (Eysenck, 1967).
The focus of this chapter is on recent developments in classical conditioning
theory and research, but it will also consider what relevance they might have to
the study of individual differences. It will attempt to bridge the gap which
currently exists between conditioning theories per se and factors of personality,
and the strategy adopted will be that of fitting personality within the wider
perspective of conditioning research. In the past, they have not been integrated
since their bases of reference are very different. Eysenck's theory is centered
on personality which, he postulates, influences sensory, perceptual, and
performance measures, including conditioning, as well as real-life (socio-
psychiatric) behavior. Conditioning theories exemplify a simple form of
associative learning, their main preoccupation being with the circumstances
under which learning occurs, the content of the learning ("what is learned")
and how learning is manifest in performance. Developments in conditioning
theory have been rapid and it is the case that Eysenck's theory was applied to
an earlier and simpler model of conditioning, far removed from the
sophisticated paradigms explored today. Cognitive interests have opened up
new questions and offer new techniques for examining them.
The main lines of research in contemporary classical conditioning which will
be discussed are: first, a highly effective associative theory based on animal
research; a separate and highly successful set of studies tracing the essential
circuits of the brain involved in classical conditioning; and an ongoing debate
concerning the contribution of "mechanistic" associative and higher "cogni-
tive" factors to human conditioning. Eysenck's theory stands much as it was in
1967, and although vastly more is now known about factors influencing
individual differences, and about the contribution of genetic, physiological,
biochemical, and psychological factors to personality, this knowledge has not
been used to make new and specific predictions about personality and
conditioning.
This chapter will review animal studies, both behavioral and physiological,
and then human research into cognitive factors. It will consider the various
explanatory mechanisms in contemporary use and their possible contribution
towards a better understanding of human conditioning. In the cognitive era
which replaced behaviorism, the mechanistic implications of the Hullian
approach made classical conditioning seem irrelevant to human learning. Yet
 Classical conditioning and the role of personality 341

there are signs that this neglect is being replaced by serious attempts to
integrate human, animal, and brain research, and by an awareness of infor-
mative interactions between apparently diverse interests. Human classical
conditioning is hopefully emerging from a period of isolation. It is, of course, to
Eysenck's credit that he has continuously supported this kind of research, and
this chapter will begin with a summary of his 1967 theory and research carried
out in his laboratory since that date.

2. EYSENCK'S THEORY

This was developed against a background of physiological and behavioral

theorizing. Pavlov's model of the cortex, consistent with that of his
contemporaries, invoked concepts of cortical excitation and inhibition, and
their balance. The typology which evolved consisted of three fundamental
properties of the nervous system: strength, equilibrium, and motility, these
features being reflected in the speed, stability, and flexibility of conditioned
response (CR) formation. This typology was subsequently extended by Teplov
and Nebylitsin. They proposed a close dependence between the strength of the
nervous system, defined by Pavlov as the ability of cortical cells to work under
intense or prolonged stimulation, and sensitivity as measured by sensory
thresholds. The "weak" nervous system type is relatively sensitive to low
intensity stimulation but at high intensities is likely to show a reduction of
function-labeled transmarginal inhibition. The "strong" nervous system type
has more "functional endurance" or capacity for continued response to in-
creasing intensity of stimulation (Nebylitsin and Gray, 1972). The many
insights provided by these researchers have been somewhat overshadowed by
their use of outdated and overly general concepts of central nervous system
functioning.
However, in the 1950s it was reasonable to refer to Pavlovian concepts of
excitation, inhibition, and the balance between them, and they were used by
Eysenck as the main constructs to explain the physiological basis of E-I.
Together with Hull's reactive inhibition, they created a theoretical blend which
allowed Eysenck to make strong predictions about classical conditioning
performance: introverts were identified with the weak nervous system, and
hence with sensitivity to stimuli, that is, low sensory thresholds. They behave as
if they amplify stimulation. They are also slow to build up Hullian reactive
inhibition, a decrement in responding applied both to massed stimulus input as
well as to motor "work" decrement. (The need to consider sensory input and
motor output as separate factors will be further discussed in relation to
"cognitive"- and "motor"-type inhibitions in contemporary research.)
342 Further Eysenckian interests

There has only been one other, relatively short-lived, attempt to bring the
study of individual differences within a theoretical framework, and that was
carried out by Spence, using the Hullian hypothetico-deductive formula
(Spence, 1964). He postulated that individuals high in anxiety drive would
condition more rapidly. Anxiety was assessed by means of the Taylor Manifest
Anxiety Scale, and unconditioned stimulus (UCS) intensity was the variable
used to manipulate anxiety levels. A controversy arose between Eysenck and
Spence, Eysenck emphasizing inhibition and Spence anxiety as the major
determinants of conditioning. Some resolution of this controversy was reached
through the realization that Eysenck had selected experimental conditions
conducive to inhibition, while Spence had selected those designed to arouse
anxiety. The details are reviewed elsewhere and are perhaps of historical rather
than theoretical interest today (Levey & Martin, 1981); they serve, however, as
a further reminder of the potency of stimulus factors in the study of individual
differences in conditioning.
In 1967 Eysenck modified his theory and the terms inhibition and excitation
were now reconsidered in terms of arousal. It was postulated that extraverts
have a chronically low level of cortical arousal and are more susceptible to
arousal decrement than introverts, who have a relatively high level of cortical
arousal associated with a lesser susceptibility to arousal decrements. The
notions of transmarginal inhibition and a curvilinear relationship between
performance and arousal were retained: under moderately stimulating
conditions introverts were predicted to perform better than extraverts, and
less well under high stimulation or stress. Three conditions were stated to
influence excitation/inhibition or arousal in the conditioning situation: high
versus low UCS intensity, continuous versus partial reinforcement, and short
versus long interstimulus intervals (ISIs).
Eysenck proposed that neuroticism (N) depends on the functioning of the
visceral brain and claimed that differences between those high and low on N
could be accounted for in terms of differential thresholds for hypothalamic
activity and differences in responsiveness of the sympathetic nervous system,
high N scores being associated with high responsivity. Thus a distinction was
made between cortical arousal (related to E-I) and autonomic activation
related to N. Less research has been devoted to conditioning and N than to
Introversion-Extraversion. In part this is due to difficulties in realizing
stimulus conditions that are sufficiently stressful to elicit individual differences
in activation, and in part to the factor of response specificity such that
individuals differ in the autonomic response systems which are activated under
stress. Fahrenberg's research investigating a wide range of response systems
under a variety of experimental conditions illustrates the factor of response
specificity in the apparent absence of generalized sympathetic responsivity in
high N people (Fahrenberg, 1987). The dimensional picture has become more
complex with the development of Psychoticism and Impulsivity scales.
 Classical conditioning and the role of personality 343

An eyelid conditioning study by Levey (1972), described in Eysenck and

Levey (1972) illustrates the approach outlined above. Subjects were tested on
the Eysenck Personality Inventory (EPI) [the scale which succeeded the older
Maudsley Personality Inventory (MPI) and preceded the Eysenck Personality
Questionnaire (EPQ)], and were assigned at random to combinations of three
experimental conditions, viz. high versus low UCS intensity, continuous versus
partial reinforcement, and short versus long ISIs. The chief interest of the
results lay in the comparison of the combined effects of these three
experimental conditions conducive to inhibition or arousal. For the
inhibition-producing factors the extravert group virtually failed to condition.
Under high arousal conditions, the extraverts gave more CRs than introverts, a
result interpreted in terms of transmarginal inhibition: at some point above
optimal stimulation the weak nervous system of the introvert becomes
susceptible to protective inhibition and performance declines.
A further study (Jones, 1975) using UCSs of very low intensity also found
that under the low arousal conditions, extraverts showed lower levels of
responding. This study also reported differences in topographical measures
such as peak latency and amplitude, the extraverts showing lower amplitudes of
responding and shorter peak latencies, consistent with an inhibited or under-
aroused performance. During the 1960s, Levey and Martin had expanded the
assessment of conditioning performance beyond simple counts of CRs to
consider topographical measures, in particular the placing of the CR in
relation to the UCR (Martin and Levey, 1965; Martin and Levey, 1969). Over
the course of learning, the CR changes its shape and frequently blends with the
UCR to form a single, smooth response with a peak latency close to the
occurrence of the UCS. It was argued that individual elements such as onset
latency, rise time, amplitude, and peak latency must be integrated to ensure
appropriate placement and blending of the CR with the UCR. These measures
were only poorly correlated with CR frequency, opening up the possibility that
simply to count CR frequency was an inadequate account of performance. The
observation that different measures of conditioning are poorly correlated had
been made earlier by Humphreys (1943) but such findings are at odds with
behavioral theories (both old and new), which assume that all such CR
measures "index" the strength of some hypothetical association.
Eysenck's emphasis on the importance of stimulus conditions in studying
individual differences in classical conditioning has been well justified, and the
results of several studies support his predictions concerning E-I and
conditioning but with some qualification, in particular concerning the role of
Impulsivity. The Levey study using the EPI noted that an ad hoc nine-item
impulsivity scale drawn from the EPI was able to account for all the differences
between extraverts and introverts under extreme stimulus conditions. Another
study (Barratt, 1971) emphasized the effects of impulsivity (using a
questionnaire of his own design), this time crossed with anxiety. EEG
344 Further Eysencldan interests

recordings showed that the Hi imp/Low anx Ss were less aroused at the
moment of stimulus presentation and gave significantly fewer CRs in
acquisition, while the greatest number were given by Lo imp/High anx Ss.
Although the group membership used by Barratt has been explored by Gray in
his reformulation of Eysenck's dimensions (Gray, 1981), further research is
still ongoing in the attempt to clarify the nature and role of impulsivity.
In summary, the contribution of the personality dimensions to human eyelid
conditioning is not entirely clear-cut. Investigations have to consider the
interactions between dimensions, the possible role of impulsivity, and the
subsequently developed Psychoticism (P) factor. One of the later studies in the
series of eyelid conditioning experiments involved examining the interaction of
P and E, again varying UCS intensity. Unlike the other factors, there is no firm
basis for predicting the conditioning performance of high versus low P subjects.
A large-scale eyelid conditioning study with 160 subjects investigated the
factors of P and E crossed with two levels of UCS (airpuff) intensity, one
strong (7 psi) and one weak (2 psi) (Frcka, Beyts, Levey, & Martin, 1983). Sex
and blink threshold were also included as factors. The importance of UCS
intensity was supported by a highly significant P x E x UCS intensity
interaction which upon further examination was shown to be entirely due to
the low UCS intensity condition: lowP/lowE and HighP/highE Ss gave more
CRs than the HighP/lowE and LowP/HighE groups. Differences between
groups were less pronounced under the high UCS intensity condition.
Testing of the main effect of E within the low P group showed that E had a
significant effect on CR frequency in the predicted direction. Thus, provided
the P score was low, the predicted high CR frequency for low E Ss and low CR
frequency for high E Ss, was observed. However, within the high P group the
situation was reversed: high E Ss performed relatively well whereas low E Ss
performed relatively poorly.
These results raise the question of whether the HighP/HighE and LowP/
LowE differences might be explained in terms of Impulsivity, the former group
being high and the latter low. Although Levey and Barratt had both implicated
Impulsivity in their findings, their impulsivity scales were different from
current Eysenckian versions which now define two factorially distinct aspects of
impulsivity, one aligning with P and named Impulsivity (Imp), and the other
with E and named Venturesomeness (Vent). Imp correlates mainly with P and
N, and modestly with E, while Vent correlates mainly with E and modestly with
P. Frcka collapsed her experimental groups to form new groups of high and
low Imp, and analyzed these with factors of UCS intensity and sex. No
significant main effect was observed on CR frequency or CR amplitude.
However, the trial number of first CR revealed a significant difference, high
Imp Ss starting to respond later than low Imp Ss. A significant three-way
interaction of Imp x UCS intensity x sex was also significant for trial number
of first CR. All groups started to respond earlier at the higher UCS intensity,
 Classical conditioning and the role of personality 345

but low Imp females under the high UCS intensity were the earliest group to
respond and high Imp males under the low UCS intensity were the latest to
respond (Frcka and Martin, 1987).
Such a result is at variance with the conception of impulsive people as
responding more quickly to stimuli and/or failing to withhold a response that
will lead to punishment. Research continues to analyze the various items
comprising impulsivity and its behavioral components, especially in relation to
clinical impulsive behavior. Recent studies with Imp and P have employed a
range of tests and laboratory tasks which implicate styles of motor responding
and styles of information processing, the latter based on the hypothesis of
deficits in selective attention mechanisms in high P subjects (see later section).
Although, as indicated, there have been few specific predictions about P and
conditioning, there have been suggestions that high P people are insensitive to
punishment. This would be consistent with reports linking P to psycho-
pathology and criminal behavior. A further study was conducted which used
paraorbital shock as the UCS. It comprised a factorial design with two levels
each of P, E, UCS intensity, and blink threshold (Beyts, Frcka, & Martin,
1983). Results showed a significant main effect for P, high P Ss conditioning at
a lower level. Further, in combination with a high blink threshold, high P
subjects showed a marked reduction in UCR amplitude over trials. Such
results add weight to the view that different UCSs might interact with the
personality dimensions, as for example in Gray's (1981) hypothesis of
differential susceptibility to reward and punishment. This hypothesis also
implicates anxiety. Correlations between anxiety ratings and all measures of
conditioning were, however, uniformly nonsignificant. This result may be
attributable to the very low levels of shock UCS intensity used: UCR frequency
of responding dropped to about 55% on the last block of acquisition trials.
Both the Levey (1972) and the Frcka et al. (1983) study reported highly
significant effects for blink threshold to the airpuff on CR frequency, low
thresholds resulting in higher levels of CR frequency. The Beyts et al. (1983)
study also assessed blink thresholds to the paraorbital shock and similarly
found that Ss with low blink thresholds gave more CRs, showed less UCR
habituation and perceived the shock as more intense.
It is evident that attempts to examine individual differences in a simple
conditioning situation have generated a range of significant findings which rest
on an early inhibition/arousal theory. Although Eysenck's predictions relating
to E-I have been upheld, little experimental work has been carried out with the
N factor, and research with P has provided problematic data. Theoretical terms
such as inhibition and excitation are no longer subject to the rigorous opera-
tional definitions of the Hullian era and are used in different forms today as
explanatory concepts. The term arousal although still widely used has defied all
attempts to describe or define it. Yet there are repeated suggestions that it
affects or interacts with sensory thresholds, response thresholds, attention,
346 Further Eysenckian interests

vigilance, and orienting. There has been no systematic program to examine

these factors although they are all relevant to conditioning. Response
thresholds are demonstrably important, habituation of the UCR may occur,
attention fluctuates over trials, and some subjects even become drowsy at the
end of the testing session. All of these factors may influence conditioning
processes and implicate individual differences. Eysenck has frequently
acknowledged the weakness of the "cortical arousal" theory of personality
but would argue that its use has led to many new discoveries in experimental,
social, occupational, and clinical psychology. What is being examined here,
however, is whether there has been any progressive development specific to
conditioning and personality. Before considering this in detail it is appropriate
to examine what advances have been made in classical conditioning in other
areas during recent years.

3. CONTEMPORARY ANIMAL CONDITIONING THEORY

Classical conditioning is associative but also cognitive in that it attributes to the

organism a sensitivity to informational relations among events. The attempt to
maintain human conditioning within a theoretical framework based on animal
research which specified the growth of excitatory and inhibitory potentials as
the determinants of classical conditioning was an important feature particularly
of the Spence but also the Eysenckian approaches. Any cognitive involvement
(knowledge, expectancies, verbalizable reports) was largely ignored or "hand-
led" by instructions ("just let your reactions take care of themselves").
Research into human conditioning outside animal associative theories began to
take cognitive factors seriously, to the extent that in 1974 Brewer put forward
the thesis that all human conditioning could be considered in cognitive terms.
This widely cited review led to a further division between those who considered
classical conditioning largely in terms of inhibitory and excitatory associations
and those who emphasized the importance of verbalizable expectations and
knowledge. Although interest in Hullian theory declined, the immensely
influential conditioning theory developed by Rescorla and Wagner (1972) has
its roots in association theory, as had Hull's. But whereas the traditional view
of conditioning was of a passive, automatic process, often implying an equal
access of CSs to association with the UCS, the newer view stresses the many
factors which can determine the associability of a CS, its stimulus history, the
context, and competition with other concurrent stimuli, for example. It
emphasizes the importance of a discrepancy between the actual state of world
events and the organism's representation of that state: only surprising or
unexpected reinforcers will receive sufficient processing to sustain condition-
ing. It stresses that the growth of excitatory and inhibitory processes can lead to
 Classical conditioning and the role of personality 347

sophisticated predictive or informational relations between potential CSs and

the UCS.
A significant facet of the theory is its emphasis on context, that is, stimuli
which are concurrent when conditioning occurs. Pavlovian learning involves a
mechanism whereby the evocation of CRs is not simply dependent on the CS-
UCS association but on additional stimuli which can modulate the expression
of the simple association. On the basis of these occasion-setting stimuli or
environmental cues, the organism comes to discriminate those occasions on
which (for example) the CS will and will not be reinforced. Thus, within animal
conditioning theory, the old view of conditioning as the establishment of new
reflexes or the strengthening of S-R connections has given way to a view of
conditioning as the acquisition of knowledge, knowledge which may not be
immediately apparent in changes in behavior and which may require specific
designs to reveal such changes. Further, when animals learn an association, this
is achieved by the establishment of an association between some central
representation of the two events.
This development of animal conditioning theory has been responsible for
ingenious methodology in designs which analyze the excitatory/inhibitory
nature of the associations between stimuli, and has led to the view that such
associations are not simply binary between two events, but can become
hierarchical in their structure. Associations among some pairs of items yield
new entities that can themselves enter into further associations. This has
immensely extended the power of the earlier simpler associative theory
towards one which can build complex performances based on elementary
mechanisms. There are similarities in outlook between this and connectionistic
theories which appeal to multiple associations interacting to produce complex
outputs.

4. CONTEMPORARY NEUROPHYSIOLOGICAL RESEARCH

Parallel with these developments has been the research into brain substrates of
simple classical conditioning. Thompson and his colleagues have used the
nictitating membrane extension response (NMR) in the rabbit as a model
paradigm to identify the essential neurophysiological events in the brain. A
critical aspect of this research is circuit analysis: tracing the neuronal pathways
from the CS and UCS channels to the motor neurons. As the essential
pathways and structures are defined it becomes possible to localize and analyze
cellular mechanisms underlying learning. A distinction must be made between
the learned response circuit and the memory trace itself, the latter seemingly
localized in multiple traces in the cerebellar cortex (Thompson, 1991). Studies
to date have revealed the importance of two higher brain systems that become
engaged in this paradigm: the cerebellum and the hippocampus.
348 Further Eysencldan interests

Substantial evidence indicates that the essential memory trace for the
eyeblink CR is formed and stored within a localized region of cerebellum
encompassing the interpositus and dentate nuclei and overlying regions of
cerebellar cortex. Lesions in this area, ipsilateral to the trained eye, cause
abolition of the behavioral CR but not the UCR. The role of the hippocampus
is less clear. Two important observations have been that hippocampal unit
activity during classical conditioning is positively correlated with the
topography of the overt behavioral CR and that during acquisition,
increments in hippocampal unit activity precede the acquisition of the CR
and form a predictive model of the amplitude-time course of the learned
behavioral CR. Hippocampectomy has little effect on the acquisition of CRs
under simple experimental conditions but may produce deficits in the
acquisition of responding under complex stimulus conditions such as discri-
mination reversals or "if-then" conditional operations.
There are also interactions between the hippocampus and cerebellum,
leading to the possibility of modulation of inputs to cerebellar areas, between
the hippocampus and the reticular formation, and between the hippocampus
and cortex. Clark, McCormick, Lavond, and Thompson (1984) found that the
hippocampal neuronal activity reflecting the topography of the rabbit NMR
disappears after cerebellar ablation, supporting the idea that CS-US associa-
tions are stored in cerebellar areas and not in the hippocampus.
Thompson's research has in the main employed the rabbit NMR. Other
workers have been interested in the neuronal regions underlying a variety of
response systems. There have been frequent distinctions between diffuse
preparatory CRs (typically autonomic), and consummatory CRs such as eyelid
closure which involve learning precise, adaptive CRs that deal specifically with
the UCS. Studies of cardiovascular conditioning suggest that the amygdala and
cingulate cortex seem to be involved in the mediation of associative processes
underlying autonomic learning. Lesions in these areas selectively affect
classically conditioned bradycardia responses in the rabbit ("selectively" in
that they have no effect on other classically conditioned responses such as the
NMR) (Jarrell et al., 1986). These findings support the view that the heart rate
and NMR-conditioned responses are mediated through different neural
pathways. As yet this research into specific motor conditioning and diffuse
autonomic conditioning remains unintegrated and it is uncertain whether we
should consider conditioning of these different response systems as inde-
pendent, or related to one another. One proposal is that autonomic changes
involve attention and/or emotion which are necessary for somatomotor
responses to be conditioned.
Another memory trace system has been labeled "cognitive" or "declarative"
(Squire, 1988). Squire has proposed that declarative memory may be a new
development in evolution that corresponds with the elaboration of the
hippocampus and other higher brain systems. Damage to diencephalic or
 Classical conditioning and the role of personality 349

hippocampal-temporal structures in humans results in marked deficits in the

formation of declarative but not procedural memory. A broad range of cont-
emporary neuropsychological research into memory functions parallels
conditioning research, but has made more progress in integrating findings
from clinical data, particularly from amnesic human patients, with lesion
studies in monkeys and rats. Squire (1992) argues that the hippocampus is
essential for a particular kind of memory function which he labels as
declarative, but that the amygdala is not part of this functional system. Work in
both monkeys and rats suggests that the amygdala is important for other
functions, including the acquisition of conditioned fear and the establishment
of affective significance for neutral stimuli (LeDoux, 1992).
Simple classical conditioning paradigms would seem to be an instance of
procedural memory. However, when tasks are made more complex it is
possible that declarative memory becomes involved and that the hippocampus
then plays a more critical role. In the context of Squire's theory it might be that
declarative memory has developed from the more ancient procedural memory
system, the latter involving the cerebellum. If so, then one might expect the
hippocampal system to become engaged in all learning paradigms, even though
it is only essential in situations that require some aspect of declarative memory.
It is interesting and convenient that the human conditioned eyelid response
largely mirrors that of the rabbit nictitating membrane. The clear parallels
between rabbit and human findings suggest that studies involving brain-damaged
humans might provide fruitful data for comparison with animal lesion studies.
Recent research by Daum and associates shows that acquisition of the CR under
very simple stimulus conditions is severely impaired in patients with cerebellar
lesions (Daum et al., 1993) but that there is no effect on conditioned electro-
dermal responses. Both cerebellar patients and controls could accurately
verbalize the relevant stimulus contingencies under the simple stimulus schedule
employed, a finding in line with earlier results showing that verbalizable
awareness did not significantly aid discrimination of temporal lobectomy patients
(Daum, Shannon, Polkey, & Gray, 1991). This latter study showed that CR
acquisition is disrupted in patients with temporal lobe-hippocampal regions if the
conditioning situation extends beyond simple tone-airpuff associations, as in the
acquisition of conditional discriminations. The potential significance of animal
conditioning studies for the role of individual differences in human conditioning
is enormous, but must rest on careful comparison of observable differences in
performance as they relate to known brain functions.

5. THE COGNITIVE-BEHAVIORAL DEBATE

By contrast with the more directed efforts of animal researchers, recent human
research has been motivated by a number of different goals. One is to see
350 Further Eysencldan interests

whether the most influential animal conditioning theory (Rescorla & Wagner,
1972) can be applied to humans. The advantages of a theory which can be
generally applied are great; Hull, Spence, and Eysenck have argued for this
approach in the past, and more recently Davey (1987) has put the case for
integrating human associative learning within animal conditioning theory. His
suggestion is that human conditioners could adapt the inferential techniques
developed in animal studies to make more precise predictions about what
associations are learned and how cognitive representations of external events
mediate conditioning behavior. Davey gives examples of studies which support
his contention, a particularly interesting example being techniques of
postexperimental revaluation of the UCS to infer the nature of the CS-UCS
representation. Other studies designed to explore Kamin's blocking phenom-
enon, which relates to the associability and predictive value of a CS, have been
only moderately successful (Davey & Singh, 1988; Kimmel & Bevill, 1996;
Lovibond, Siddle, & Bond, 1988; Martin & Levey, 1991). Only one has been
directly concerned with individual differences, and this reported no systematic
relation between the blocking effect and several measures of psychoticism
(Jones, Gray, & Hemsley, 1990).
A characteristic of the Rescorla-Wagner theory is its emphasis on stimulus-
stimulus (S-S) learning to create a detailed inter-event structure of the world.
The animal does not randomly form associations between any two stimuli that
happen to occur, rather it is an information seeker of logical and perceptual
relations among events. Such a view has been extended to human research,
which points to humans as learning about the "causal structure" of the world,
the idea being that what an animal learns during CS-UCS pairing is analogous
to what humans learn during the pairing of cause and effect in causal learning
(Dickinson, 1980; Shanks & Dickinson, 1987).
While contemporary animal and neuroanatomical researchers often discuss
cognitive contributions to conditioning they fail to throw any light on an issue
which many have thought to be a vital factor in human studies. Can subjects'
learning be modulated by instructions, information about, or knowledge of, the
experimental schedule? This has been an area of controversy since the earliest
studies (e.g., Hilgard & Humphreys, 1938), many of which have asked whether
information given to subjects about the stimulus schedule pre-experimentally,
or whether knowledge acquired during the experiment, could affect
performance. Many studies have demonstrated an effect, generating a debate
as to whether classical conditioning is an automatic process over which there is
no cognitive control, or whether, as argued by Brewer (1974) it is an entirely
cognitive process determined by subjects' knowledge of stimulus relationships
and their decision to respond accordingly. If the latter, a one-to-one corres-
pondence of knowledge to performance would be expected, a result never
obtained in human conditioning studies.
 Classical conditioning and the role of personality 351

The evidence (discussed below) points to a contribution from both

mechanistic and intentional processes. A critical question is how to
conceptualize their influences in terms of specific predictions from "cogni-
tive" and "automatic" sources. The term "cognitive" is of course used in
various ways. Animal researchers (e.g., Rescorla, 1988; Holland, 1993) use the
term to refer to the animal's capacity to learn stimulus relationships and to the
representation of such relations. They argue that animals learn sophisticated
predictive and informational relations among CSs and the UCS, and that
associations are formed between internal representations of the CS and UCS
which are activated by external events. The "cognitive" argument in human
studies has centered on subjects' verbalizations about stimulus relationships
and knowledge of the schedule deemed to be conscious. Advances in clarifying
the issue have been handicapped by the extreme difficulty of any agreed and
adequate method of assessing subjects' knowledge and also by the absence of
any general theoretical consensus. Cognitions have been linked to the concept
of a limited central processing system (e.g., Dawson & Schell, 1987), pre-
attentional mechanisms (Ohman, 1992), propositional expressions (Furedy &
Riley, 1987), and trial-by-trial expectancies (Furedy & Schiffmann, 1973).
Several attempts have been made to bypass the problems of self-report by
using a variety of masking techniques. Using a concurrent auditory perception
task, Dawson and colleagues have concluded that classical differential
conditioning of the electrodermal response occurs only among subjects who
become aware of stimulus relationships (Dawson & Biferno, 1973). This
conclusion has been widely accepted in electrodermal conditioning.
There is reason to suppose that autonomic and motor response systems may
exhibit differences in the degree and perhaps the manner in which they can be
controlled by cognitive processes. Certainly the literature on electrodermal
conditioning has consistently produced strong evidence arguing for the effects
of knowledge on responding, although as Furedy has pointed out, there is no
one-to-one correspondence between a subject's stated expectancy of whether
the UCS will occur on a given trial and the actual occurrence of an electro-
dermal response. Motor responding, and eyelid conditioning in particular, has
generated a particularly troublesome topic concerning the operation of
voluntary factors, so-called voluntary response "sets," "self-instructions," and
voluntary or deliberate eye closure to the CS. Analysis of CR forms suggested
to American workers that there are two categories of response, a "judged
voluntary response," which resembles blinks made under instructions to blink
to a signal, and a "true" nonvoluntary CR. Attempts to define these categories
in terms of response slope and latency failed to achieve any general consensus.
Grant, however, explored the distinction between those individuals showing
a mainly voluntary response (Vs) and those a conditioned response (Cs). He
had been moving towards an information-processing account of conditioning
by employing CSs that were not neutral lights or sounds but words and/or
352 Further Eysenckian interests

messages, e.g., color words such as "Blue" as CS+ and "Pink" as CS—; "Don't
blink" as CS+ and "Blink" as CS-. He suggested that those individuals
showing a mainly voluntary blink form (Vs) and those a mainly conditioned
response form (Cs) react differently to verbal conditioned stimuli and process
the verbal information differently. Grant postulated a trade-off between the
demands of central processing and those of response shaping. Vs have more
central processing capacity available to process the CS more analytically, and
develop a response topography that is more effective in reducing cornea!
airpuff (UCS) stimulation. Since this V response is readily available it makes
little demand on central processing capacity, hence making more available for
an extensive and elaborate analysis of the CS (Grant, 1972).
Work carried out in Eysenck's laboratory has also been concerned with the
effect of knowledge on eyelid conditioning performance (Kayata, 1987; Martin
& Levey, 1987). The evidence that knowledge affects this response modality
has been less clear-cut than in the electrodermal response system. Kayata
observed that a group who were given full information about the conditioning
schedule—a discrimination schedule mediated by an occasion-setting
stimulus—showed differential responding to the reinforced CS (CS+) versus
the nonreinforced CS (CS—), responding to the former but not to the latter.
Some subjects who were not given the information but became aware of the
schedule through exposure to it also demonstrated good differential respond-
ing. However, those who were not informed showed a normal acquisition curve
increasing significantly over trials for both CS+ and CS-. Previous studies
have not always made it clear that knowledge may significantly influence
responding under complex stimulus schedules, but that in the absence of any
knowledge, subjects will still demonstrate normal acquisition curves within
these schedules. A mechanistic process would seem to ensure that conditioning
occurs even though the response pattern may not show a precise
correspondence with the detail of the experimental schedule. This kind of
schedule has particular interest in that animal researchers have shown that the
hippocampus seems to be necessary for successful conditioning in similar
discrimination (if-then), and occasion-setting paradigms. Such schedules may
therefore provide a useful base for human neuropsychological studies to link
with animal work.
Recent work by Ohnian (e.g., Ohman, 1992) has used the technique of
backward masking to prevent conscious recognition of the conditioning stimuli,
with the aim of analyzing possible automatic mechanisms. Their work on
phobias derives from Seligman's (1971) suggestion that humans may be
evolutionarily predisposed to react more strongly to potentially threatening
objects. Ohman's approach asks whether processing of the phobic stimuli is
rooted in archaic information-processing mechanisms outside the control of
conscious intentions. If it is controlled from preattentive, unconscious in-
formation processing mechanisms then the emotional response to a phobic
 Classical conditioning and the role of personality 353

stimulus may be initiated before it is consciously recognized. The results of a

long series of electrodermal conditioning studies has led Ohman to the
conclusion that conditioned autonomic responses to fear relevant animal
stimuli can arise after a merely superficial automatic evaluation of the stimulus
occurring outside awareness.
A different aim has been pursued by Levey and Martin in their evaluative
conditioning research (Levey & Martin, 1975; Martin & Levey, 1978). Their
experiments shift attention from the learning of stimulus relationships, as
emphasized in contemporary conditioning theory, and from the overt
production of conditioned responses, to a more central aspect of condition-
ing—the positive or negative change in the evaluation of neutral stimuli when
paired with actively liked or disliked stimuli. The materials used are typically
picture postcards of unfamiliar paintings, and Ss are required to sort them into
"liked," "neutral," and "disliked." The pictures are then presented tachisto-
scopically, in pairs: neutral-liked; neutral—disliked; and neutral-neutral, for
example. After exposure, Ss rate the pictures on a numerical scale of "liked" to
"disliked." Results show significant changes in the ratings of neutral stimuli,
which increase towards "liked" or "disliked," depending on the pairing. It is
hypothesized that what is primarily learned is some form of central evaluative
state, triggered by an evaluative response to a salient stimulus. It leaves open
the nature of the overt response: a given positive or negative evaluation will
result in different decision or action patterns on different occasions.
A requirement of the experiment is that Ss make evaluative and not
cognitive judgments. Although Ss are routinely told that the experiment is
about simple, spontaneous likes and dislikes, many respond to the form or
structure of the picture rather than making a direct evaluation. A set of
experiments was therefore designed to test the observation that Ss, who
respond more on the basis of an immediate "feeling" response, condition
better than those who respond on the basis of cognitive judgment. A
comparison of these "feeling" and "cognitive" groups, identified by a brief
questionnaire, showed the conditioning effect more clearly in the former
group. A further experiment selected extreme groups on the dimension of E-I
and showed that extraverts conditioned better, in agreement with the
observation that extraverts are more likely to give a direct "feeling" response
than introverts, who are more likely to adopt an analytical, cognitive approach
to the materials (Martin & Levey, 1978).

6. BEHAVIOR THERAPY

The most important application of Eysenck's theory of classical conditioning

has been to behavior therapy. His theory states that the different types of
neurotic illness arise through Pavlovian conditioning, and can be eliminated
354 Further Eysenckian interests

through a process of Pavlovian extinction (Eysenck, 1979). Much was made of

the predictive value of conditioning theory in generating procedures for
producing extinction of maladaptive responses. Exposure, flooding with
response prevention, desensitization, and modeling were all found to be
effective. Although the emphasis was often on behavioral responses, the
contribution of emotional factors was well to the fore. Mowrer's two-factor
theory, for example, postulated that emotions play a central, indispensable role
in behavior performance change. Emotions were what is learned and were
essential to a theory of learning, behavior modification, and control. Traumatic
avoidance learning and two-process theory in general formed the conceptual
framework for much of the early behavior therapy and research; symptoms and
defensive maneuvers could be viewed as equivalent to avoidance behavior and
conditioned fear/anxiety reactions could be altered by deconditioning techni-
ques. These certitudes on the central role of conditioning in clinical anxiety
were soon questioned by cognitive therapists making strong claims for focusing
on cognitive events in therapy. Over the years the emphasis has shifted towards
the use of cognitive components in therapy, sometimes to the exclusion of
psychophysiological and behavioral assessments.
The interaction between behavioral conditioning theories and therapy is
pretty well nonexistent today. Although contemporary behavior theory
provides a more liberal model of conditioning than the traditional one, it is
also a more complex and uncertain one. Whereas the behavior therapists of the
previous generation could treat conditioning as a simple and well-understood
phenomenon, they now find in the literature a plethora of effects competing
for a variety of theories; a stimulus can be a conditional excitor, a conditional
inhibitor, or even an occasion setter; it can have a high associative strength but
low associability, or any other combination of these properties (Dickinson,
1987). Eysenck, of course, has always moved with certainty among the
competing claims of behaviorists and cognitivists, and has been ever willing to
consider new evidence, whether from cognitive sources or, for example, work
on neurohormones showing they can have a profound modulating influence on
resistance to extinction (Eysenck, 1987). Behavior therapists have moved more
cautiously, frequently maintaining that their approaches do not depend upon
any detailed theoretical understanding of laboratory paradigms. There are not
many bridges between contemporary conditioning theories and therapy; an
exception is the work of Bouton (1988) who considers the importance of
environmental context in fear conditioning and extinction.
Bouton reviews evidence that extinction does not produce a permanent
unlearning of fear. A valid alternative is that extinction yields a stimulus with
reduced but quite volatile power. Clinically, recent exposure to a traumatic
stimulus or stressful experience may not be sufficient by itself to reinstate an
extinguished fear. Instead, fear returns primarily when the extinguished
stimulus is encountered in a fearful context. It is as if the conditioned context
 Classical conditioning and the role of personality 355

arouses an expectancy of the UCS and it is this expectancy that reinstates fear.
The argument is derived from animal conditioning theory and uses its
explanatory concepts. Contexts can enter into direct associations with CSs but
contexts may also select among and signal CS-UCS associations; for example
"the tone is dangerous in context A" rather than "context A is dangerous."
The point has significant theoretical implications. If contexts sometimes select
from a range of associations instead of simply entering into associations
themselves then a new or different set of associative principles may apply
(Bouton, 1988).
This kind of account obviously involves a more complex structure than is
required by traditional associative models. Modulatory mechanisms such as
occasion-setters or contexts show how animals might retain a more complete
account of their conditioning experiences than is afforded by simple
associations alone (for a detailed argument see Holland, 1992, and Swartz-
entruber, 1995). Such accounts are concerned with the nature of learned
excitatory/inhibitory associations and a theoretical structure which can
encompass far more complex and hierarchical associations than the simple
binary models of earlier theories. They are more interesting because of their
vastly greater explanatory potential.

7. THEORETICAL MATTERS AND APPLICATIONS TO PERSONALITY

This section will consider trends in contemporary classical conditioning which

are likely to influence future human research, and will also examine some of
the explanatory concepts which are employed.
It seems clear that the impetus driving current brain research will continue,
and that it will interact further with animal conditioning studies in clarifying
the relationship between different brain structures and types of experimental
schedules. Human studies are urgently needed here. Daum and her colleagues
have started work in this direction (Daum et al. 1991, 1993) and further
exploration is obviously desirable. So far as individual differences are con-
cerned, there is little experimental evidence for or against Eysenck's postulates
about the physiological underpinnings of E-I and N. Brain research does,
however, seem to question the generality of the conditioning process. Eysenck
has assumed a factor of conditionability that transcends specific response
systems. This looks decidedly unlikely as the neural circuits underlying at least
the motor and the autonomic systems seem to be different.
The future interaction of animal conditioning theory and human studies is
less clear. For the most part, animal conditioning theories are theories of
associative learning, and this has produced a tightly constrained experimental
and theoretical approach centered on the formation of excitatory and
inhibitory links between stimuli and their representations. Human researchers
356 Further Eysenckian interests

have not yet been able to fit comfortably into this purely associationist
framework because of the evidence on the significant effects of knowledge and
verbalizable awareness on human conditioning performance. These effects are
difficult to incorporate within any theoretical framework and have led to
diverse explanations. But there are additional considerations to take into
account. Although today's emphasis is on cognitive issues, human research is
also concerned with the development of conditioned affective and emotional
responses. Its close involvement with the behavior therapy movement
necessarily focuses on the importance of the extinction process, as Eysenck
recognized in his 1979 conditioning theory of neurosis.
Much of the human research that has been described has been concerned
with acquisition rather than with extinction. Yet there are pieces of evidence
suggesting that its neglect has been unwarranted. Most studies have shown a
quite poor correlation between acquisition and extinction and it is by no means
clear that high levels of conditioned responding are reflected in extinction. In
the electrodermal studies of phobias it is frequently observed that differences
in conditioning between fear-relevant and nonfear-relevant stimuli are more
likely to be observed in enhanced resistance to extinction than in acquisition,
and that this effect is particularly evident for fear stimuli of an evolutionary
origin. This persistence of responding in extinction occurs in spite of the
extinguished verbal expectancy report (Schell, Dawson, & Marinkovic, 1991).
What makes the extinction process particularly interesting is its complicated
relationship with cognitive and associationist factors: even behaviorists such as
Spence recognized that cognitive factors might be influential here, and several
studies have demonstrated that when subjects are told after acquisition that no
more UCSs will occur, conditioned responding is dramatically reduced. From
an associationistic perspective, animal studies suggest that the various
extinction treatments which are typically applied may partly reverse but not
eliminate the CS-UCS association, and they even raise the possibility that it
may be fully preserved. If fully preserved, what produces the apparent
decrement in responding? Rescorla (1996) speculates on the possibility that
some inhibitory process is superimposed, possibly between the CS and its
associated response.
Clinical treatments must deal with the acquisition and extinction of
emotional responses mainly developed under real-life situations of threat.
Within the typical human laboratory procedure the stimuli employed are mild,
not particularly noxious, and certainly not life-threatening. When conditioning
occurs in situations of real trauma there is little likelihood that subjects will
immediately engage in hypothesis formation, make prepositional statements,
or adopt conscious strategies of responding, although these may be made
subsequently. In such circumstances, levels of arousal and emotion must be
 Classical conditioning and the role of personality 357

high, and action must be prompt. This highlights the point that a major
component of classical conditioning, not to be ignored, is that it provides the
mechanism by which short-term emergency consequences can be anticipated.
However, cognitively oriented clinical research is concerned more with
attention than arousal. A classic and unresolved question concerns the
relationship between arousal and attention, and there is relatively little in the
conditioning or cognitive literature which throws much light on this topic.
Some ongoing research is examining the relationship between anxiety and
attentional biases on the hypothesis that individual differences in anxiety levels
are related to the functioning of the attentional system. Again, it is probably
difficult to ensure high arousal levels within typical cognitive laboratory
paradigms. Another dominant issue is whether selective attention is automatic
or controlled. Ohman's work on the acquisition of phobias is of some relevance
in that he postulates that there may be attributes of fear-relevant stimuli that
serve as automatic attention triggers and he appeals to data on event-related
potentials (ERP) to elucidate the issue. Hackley (1993) reviews hypotheses
that have been offered, ranging from a strong automaticity theory according to
which "basic, obligatory, processing of the physical features of an auditory
stimulus is unaffected by the direction of attention" (Naatanen, 1988) to the
peripheral gating theory proposed by Hernandez-Peon, Scherer, and Jouvet
(1956), according to which descending neural pathways are capable of filtering
out irrelevant auditory stimuli when the organism is attending to stimuli in
another modality. Hackley's conclusion, based on ERP data, is that they do not
support either the strong automaticity or the peripheral gating theory, and he
offers an intermediate interpretation. According to this, sensory processes are
obligatory and invariant with attention at peripheral and brain-stem levels;
then, beginning at forebrain levels, i.e. thalamus and cortex, there is transition
from full to partial automaticity.
It is perhaps surprising that concepts of inhibition and excitation have
emerged as descriptive concepts in the cognitive literature; although they are
hardly new, they are being applied in new ways, as, for example in "cognitive
inhibition." The technique of negative priming, for example, has been used to
examine the proposition that deficits of cognitive inhibition exist preattentively.
Briefly, this paradigm involves the re-presentation and naming of a target item
which has been presented as the distractor to be ignored on a preceding trial.
Beech, Bayliss, Smithson, and Claridge, (1989) used three different stimulus
presentation speeds on groups of high and low schizotype Ss and found that
the reduced negative priming effect in high schizotypes was confined to the
fastest speeds. They concluded that this was because at the fastest presentation
rates the stimuli were acted upon primarily by automatic processes of selective
attention.
358 Further Eysenckian interests

Deficits in attention have been examined in high- and low-P individuals

using a go/no go discrimination task involving errors of commission and
omission. Bullen and Hemsley (1984), and Stavidrou and Furnham (1996)
found that high P scorers show reduced "cognitive inhibition," suggesting
impaired selective mechanisms. Such a deficit implies inability to inhibit
irrelevant information with the result that many unrelated ideas become
interconnected, a "widening of the associative horizon." An excitatory/
inhibitory view of selective attention is that actively attending to some item
or event entails both the activation of that event's representation and the
inhibition of competing representations. The combination of excitation and
inhibition embedded in this interpretation might seem to offer a basic
mechanism for fine tuning a variety of cognitive processes, but at present this
remains an ambiguous and uncertain interpretation.
Nevertheless, these paradigms and views provide new opportunities for
examining the role of attention in human conditioning research. As already
mentioned, backward masking has been incorporated into conditioning
schedules, and such research could readily incorporate individual
differences. The parameters used for backward masking need to be
established for individuals rather than for groups, since an effect of arousal/
attention seems highly likely and differences between high- and low-P subjects
might well be expected (Eysenck, 1992). ERP data could be usefully included
to indicate whether the earliest sensory analysis, which is around 15 ms for
auditory, and about 80 ms for visual stimuli, show differential effects for the
personality dimensions. Such research could also contribute to the automatic/
aware issue in human conditioning, Another potentially valuable strategy
would be to use attentional instructions to subjects, that is, to direct attention
to specific elements of the conditioning situation.
A theme which has appeared frequently in connection with individual
differences is that some people analyze stimuli in great detail while others are
less likely to engage in detailed processing but have a lower response threshold.
As Brebner and Cooper (1978) put it, the introvert is "geared to inspect" while
the extravert is "geared to respond." The notion that some people cannot
inhibit responding when it is appropriate to do so is being explored in the
clinical area of impulse-control disorder. This disorder reflects a lack of motor
control, a tendency to react quickly and without thought. It has been linked
with impulsiveness, in particular a subscale of motor impulsiveness, by
Stanford and Barratt (1992). In the context of psychopathy, Thornquist and
Zuckerman (1995) have postulated a relationship with the Impulsive Sensation
Seeking scale, which represents the (in)capacity to inhibit approach behavior in
the presence of cues associated with punishment.
Throughout the history of human conditioning there have been many studies
employing "inhibitory" or "facilitatory" instructions which have been shown to
have significant effects on response levels. Interpretation of results often
 Classical conditioning and the role of personality 359

referred to "response sets," and it was assumed that the individual could
exercise some degree of control over somato-motor and even autonomic
conditioned responses. This approach would appear to be relevant to current
thinking about the importance of exercising control over behavior, and
conditioning paradigms could be explored to examine who are the people
specifically deficient in motor control of conditioned responses, and whether
this might be due to lack of inhibition, or to the effect of very high levels of
activation/arousal. It makes an interesting supplement to the analysis of the
role of knowledge on conditioning performance, which also has a long history
in the conditioning literature. Indeed, it is one of the great assets of classical
conditioning that it plays a role in so many disciplines, has been at the center of
several major theories of behavior, and can now be expanded to encompass
cognitive interests and paradigms. It can also refer to a century of solid
research findings. Regrettably, relatively few studies in the past have explored
individual differences. Today's expanded concepts of arousal and inhibition
can build on the foundation laid down by Eysenck to open new directions for
future human conditioning research. Hopefully, this will not fail to take
individual differences into account.

REFERENCES

Barratt, E. S. (1971). Psychophysiological correlates of classical differential eyelid

conditioning among subjects selected on the basis of impulsiveness and anxiety.
Biological Psychiatry, 3, 339-346.
Beech, A. R., Baylis, G. C, Smithson, P., & Claridge, G. S. (1989). Individual
differences in schizotypy as reflected in cognitive measures of inhibition. British
Journal of Clinical Psychology, 28, 117-129.
Beyts, J., Frcka, G., & Martin, I. (1983). The influence of Psychoticism and Extraversion
on classical eyelid conditioning using a paraorbital shock UCS. Personality and
Individual Differences, 4, 275-283.
Bouton, M. E. (1988). Context and ambiguity in the extinction of emotional learning:
implications for exposure therapy. Behaviour Research and Therapy, 26, 137-149.
Brebner, J., & Cooper, C. (1978). Stimulus or response induced excitation. A
comparison of the behaviour of introverts and extraverts. Journal of Research in
Personality, 12, 306-311.
Brewer, W. F. (1974). There is no convincing evidence for operant or classical
conditioning in adult humans. In W. Weimer & D. Palermo (Eds.), Cognition and the
symbolic processes (pp. 1-42). Hillsdale, NJ: Erlbaum.
Bullen, J. G. & Hemsley, D. R. (1984). Psychoticism and visual recognition threshold.
Personality and Individual Differences, 5, 633-648.
Clark, G. A., McCormick, D. A., Lavond, D. G., & Thompson, R. F. (1984). Effects of
lesions of cerebellar nuclei on conditioned behavioral and hippocampal neuronal
responses. Brain Research, 291, 125-136.
360 Further Eysencldan interests

Daum, I., Schugens, M. M, Ackermann, H., Lutzenberger, W., Dichgans, J., &
Birbaumer, N. (1993). Classical conditioning after cerebellar lesions in humans.
Behavioral Neuroscience, 107, 748-756.
Daum, I., Shannon, S., Polkey, C. E., & Gray, J. A. (1991). Classical conditioning after
temporal lobe lesions in man: Impairment in conditional discrimination. Behavioral
Neuroscience, 105, 396-408.
Davey, G. C. L. (1987). An integration of human and animal models of Pavlovian
conditioning: Associations, cognitions and attributions. In G. Davey (Ed.), Cognitive
processes and Pavlovian conditioning in humans (pp. 83-114). Chichester, U.K.:
Wiley.
Davey, G. C. L. & Singh, J. (1988). The Kamin "blocking" effect and electrodennal
conditioning in humans. Journal of Psychophysiology, 2, 17-25.
Dawson, M. E. & Biferno, M. A. (1973). Concurrent measurement of awareness and
electrodennal classical conditioning. Journal of Experimental Psychology, 101, 82-86.
Dawson, M. E. & Schell, A. M. (1987). Human autonomic and skeletal classical
conditioning: the role of conscious cognitive factors. In G. Davey (Ed.), Cognitive
processes and Pavlovian conditioning in humans (pp. 27-55). Chichester, U.K.: Wiley.
Dickinson, A. (1980). Contemporary animal learning theory. Cambridge: Cambridge
University Press.
Dickinson, A. (1987). Animal conditioning and learning theory. In H. J. Eysenck, & I.
Martin (Eds.), Theoretical foundations of behavior therapy (pp. 57-79). New York:
Plenum Press.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Charles
Thomas.
Eysenck, H. J. (1979). The conditioning model of neurosis. The Behavioral and Brain
Sciences, 2, 155-199.
Eysenck, H. J. (1987). Behavior therapy. In Eysenck H. J. & Martin, I. (Eds.),
Theoretical foundations of behavior therapy (pp. 3-35). New York: Plenum Press.
Eysenck, H. J. (1992). The definition and measurement of Psychoticism. Personality and
Individual Differences, 13, 757-785.
Eysenck, H. J. & Levey, A. B. (1972). Conditioning, introversion-extraversion and the
strength of the nervous system. In V. D. Nebylitsyn & J. A. Gray (Eds.), Biological
bases of individual behavior (pp. 206-220). New York: Academic Press.
Fahrenberg, J. (1987). Concepts of Activation and Arousal in the theory of emotionality
(Neuroticism): A multivariate conceptualization. In J. Strelau & H. J. Eysenck
(Eds.), Personality dimensions and arousal (pp. 99-120). New York: Plenum Press.
Frcka, G. & Martin, I. (1987). Is there—or is there not—an influence of impulsiveness
on classical eyelid conditioning. Personality and Individual Differences, 8, 241-252.
Frcka, G., Beyts, J., Levey, A. B., & Martin, I. (1983). The influence of Psychoticism on
classical conditioning. Personality and Individual Differences, 4, 189-197.
Furedy, J. J. & Riley, D. M. (1987). Human Pavlovian autonomic conditioning and the
cognitive paradigm. In G. Davey (Ed.), Cognitive processes and Pavlovian conditioning
in humans (pp. 1-25). Chichester, U.K.: Wiley.
Furedy, J. J. & Schiffman, K. (1973). Concurrent measurement of autonomic and
cognitive processes in a test of the traditional discriminative control procedure for
Pavlovian electrodermal conditioning. Journal of Experimental Psychology, 100, 210-
217.
 Classical conditioning and the role of personality 361

Grant, D. A. (1972). A preliminary model for processing information conveyed by

verbal conditional stimuli. In A. H. Black & W. F. Prokasy (Eds.), Classical
conditioning H. (pp 28-63). New York: Appleton-Century-Crofts.
Gray, J. A. (1981). A critique of Eysenck's theory of Personality. In H. J. Eysenck (Ed.),
A model for Personality (pp. 246-276). Berlin: Springer.
Hackley, S. A. (1993). An evaluation of the automaticity of sensory processing using
event-related potentials and brain-stem reflexes. Psychophysiology, 30, 415-428.
Hernandez-Peon, R., Scherrer, H., & Jouvet, M. (1956). Modification of electrical
activity in cochlear nucleus during "attention" in unanesthetized cats. Science, 123,
331-332.
Hilgard, E. R. & Humphreys, L. G. (1938). The effect of supporting and antagonistic
voluntary instructions on conditioned discrimination. Journal of Experimental
Psychology, 22, 291-304.
Holland, P. C. (1992). Occasion setting in Pavlovian conditioning. In: The psychology of
learning and motivation (Vol. 28). New York: Academic Press.
Holland, P. C. (1993). Cognitive aspects of classical conditioning. Current Opinion in
Neurobiology, 3, 230-236.
Hugdahl, K. (1995). Classical conditioning: only of historical interest? Journal of
Psychophysiology, 9, 97-99.
Hull, C. L. (1943). Principles of behavior. New York: Appleton-Century-Crofts.
Humphreys, L. G. (1943). Measures of strength of conditioned eyelid responses. Journal
of General Psychology, 29, 101-111.
Jarrell, T. W., McCabe, P., Teich, A., Gentile, C. G., VanDercar, D. H., &
Schneiderman, N. (1986). Lateral subthalamic area as mediator of classically
conditioned bradycardia in rabbits. Behavioral Neuroscience, 100, 3-10.
Jones, J. (1975). Relationship between UCS intensity, rest, warning signal and
personality in eyelid conditioning. Unpublished doctoral thesis, University of
London.
Jones, S. H., Gray, J. A., & Hemsley, D. R. (1990). The Kamin blocking effect,
incidental learning and Psychoticism. British Journal of Psychology, 81, 95-110.
Kayata, L. (1987). Awareness and Human Learning. Unpublished doctoral thesis,
University of London.
Kimmel, H. D. & Bevill, M. J. (1996). Blocking and unconditioned response
diminuation in human classical autonomic conditioning. Integrative Physiological
and Behavioral Science, 31, 18-43.
LeDoux, J. (1992). Brain mechanisms of emotion and emotional learning. Current
Opinion in Neurobiology, 2, 191-197.
Levey, A. B. (1972). Eyelid conditioning, extraversion and drive: an experimental test of
two theories. Unpublished doctoral thesis, University of London.
Levey, A. B. & Martin, I. (1975). Classical conditioning of human "evaluative"
responses. Behaviour Research and Therapy, 13, 221-226.
Levey, A. B. & Martin, I. (1981). Personality and conditioning. In H. J. Eysenck (Ed.),^4
model for personality (pp. 123-168). Berlin: Springer.
Lovibond, P. F., Siddle, D. A. T., & Bond, N. (1988). Insensitivity to stimulus validity in
human Pavlovian conditioning. Quarterly Journal of Experimental Psychology, 40B,
377-410.
362 Further Eysenckian interests

Martin, I., & Levey, A. B. (1965). Efficiency of the conditioned eyelid response. Science,
150, 781-783.
Martin, I. & Levey, A. B. (1969). The genesis of the classical conditioned response.
Oxford: Pergamon Press.
Martin, I. & Levey, A. B. (1978). Evaluative conditioning. Advances in Behaviour
Research and Therapy, 1, 57-102.
Martin, I. & Levey, A. B. (1987). Knowledge, action and control. In H. J. Eysenck & I.
Martin (Eds.), Theoretical foundations of behavior therapy (pp. 133-151). New York:
Plenum Press.
Martin, I. & Levey, A. B. (1991). Blocking observed in human eyelid conditioning.
Quarterly Journal of Experimental Psychology, 43B, 233-256.
Naatanen, R. (1988). Implications of ERP data for psychological theories of attention.
Biological Psychology, 26, 117-163.
Nebylitsyn, V. D. & Gray, J. A. (1972). Biological bases of individual behavior. London:
Academic Press.
Ohman, A. (1992). Orienting and attention: preferred preattentive processing of
potentially phobic stimuli. In B. A. Campbell, R. Richardson, & H. Haynes (Eds.),
Attention and information processing in infants and adults: Perspectives from human
and animal research (pp. 263-295). Hillsdale, NJ: Erlbaum.
Rescorla, R. A. (1988). Pavlovian conditioning: It's not what you think it is. American
Psychologist, 43, 151-160.
Rescorla, R. A. (1996). Preservation of Pavlovian association through extinction.
Quarterly Journal of Experimental Psychology, 49B, 245-268.
Rescorla, R. A. & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations
in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F.
Prokasy (Eds.), Classical Conditioning II. Current research and theory (pp. 64-99).
New York: Appleton-Century-Crofts.
Schell, A. M., Dawson, M. E. & Marinkovic, K. (1991). Effects of potentially phobic
conditioned stimuli on retention, reinstatement and extinction of the skin
conductance response. Psychophysiology, 22, 140-153.
Seligman, M. E. P. (1971). Phobias and preparedness. Behaviour Therapy, 2, 307-320.
Shanks, D. R. & Dickinson, A. (1987). Associative accounts of causality judgment. In G.
H. Bower (Ed.), The psychology of learning and motivation, 21, (pp. 229-261). San
Diego, CA: Academic Press.
Spence, K. W. (1964). Anxiety (drive) level and performance in eyelid conditioning.
Psychological Bulletin, 61, 129-139.
Squire, L. R. (1988). Memory and the brain. London: Oxford University Press.
Squire, L. R. (1992). Memory and the hippocampus: A synthesis from findings with rats,
monkeys and humans. Psychological Review, 99, 195-231.
Stanford, M. S. & Barratt, E. S. (1992). Impulsivity and the multi-impulsive personality
disorder. Personality and Individual Differences, 13, 831-834.
Stavridou, A. & Furnham, A. (1996). Psychoticism, trait-creativity and attentional
mechanism of cognitive inhibition. Personality and Individual Differences, 21,143-153.
Swartzentruber, D. (1995). Modulatory mechanisms hi Pavlovian conditioning. Animal
Learning and Behavior, 23, 123-143.
Thompson, R. F. (1991). Are memory traces localized or distributed? Neuropsychologia,
29, 571-582.
 Classical conditioning and the role of personality 363

Thompson, R. F., Berger, T. W., Cegavske, C. F., Patterson, M. M., Roemer, R. A.,
Teylor, T. J. & Young, R. A. (1976). The search for the engram. American
Psychologist, 31, 29-227.
Thornquist, M. H. & Zuckerman, M. (1995). Psychopathy, passive avoidance learning
and basic dimensions of personality. Personality and Individual Differences, 19, 525-
534.
Chapter 17
Eysenck's contribution to understanding
psychopathology

G. Claridge

1. MAVERICK ORIGINS

Shortly before I was asked to write this chapter, I finished reading two recently
published biographies of the late R. D. Laing (Burston, 1996; Clay, 1996).
Although some might consider my train of association here mischievous, it
struck me that the conjunction of events was more than mildly interesting, in
giving me an opening into the task at hand. For evaluating Eysenck's
contribution to our understanding of psychopathology should surely include
some comparison with his contemporaries in the field. And, as I will show,
comparison with Laing serves a serious purpose in helping to discover
important themes for our later appreciation of Eysenck.
Laing was, of course, a very different figure: his philosophy existentialist, his
clinical approach subjective, intuitive, and skinlessly emotional; his account of
personal adaptation sociopolitical; and his interest in the contribution of
biology to psychological dysfunction minimal. Hans Eysenck stands in marked
contrast: cool, rational, reductionist, and utterly devoted to the scientific
method. Indeed, if a future alien historian sought to trace the activities and
home of the challenges to conventional psychiatry going on in the London of
the 1960s she (or he) might be bewildered: would these be found in the noisy
therapeutic chaos of Kingsley Hall or in the sound-deadened conditioning
laboratories of the Institute of Psychiatry? Nor would the accompanying
literature be of much help: the decade opened with the publication of both
Laing's The Divided Self and the first edition of Eysenck's Handbook of
Abnormal Psychology.
The co-existence in the same era of these two very disparate approaches to
psychopathology says something about the psychiatric climate at the time.
Psychiatry then was a weak, somewhat pathetic, medical specialty, still
desperately seeking professional respect, as well as plausible ways of
explaining and treating the disorders for which it was mostly responsible.
Such states of affairs have advantages, of course, and in this case it exposed the
 Eysenck's contribution to understanding psychopathology 365

psychiatric establishment to healthy criticism, both from outside disciplines

and from within its own ranks (Laing, like Thomas Szasz, his North American
equivalent—if not his political ally—was medically trained).
It is tempting to conclude that this is all that needs to be said about the
simultaneous emergence of two such contrary thinkers as Laing and Eysenck:
they were merely unrelated rebels, thrown up in the flux of a psychiatry,
psychology, and (in Laing's case) whole section of society struggling to find
ways of construing individuality. But, as noted elsewhere (Claridge, 1990), this
misses one fascinating similarity between the two men's ideas; a parallel that is
significant, both historically and theoretically. I am referring to the fact that the
starting-point for Eysenck's and Laing's analyses of psychopathology was
identical: both proposed that there is an inextricable connection between
normality and disorder, between illness and health—such that, in order to
understand one, it is necessary to comprehend the other. The way in which
each of them developed this notion was, of course, quite different: statistical
and nomothetic, in one case, clinical and idiographic, in the other. The deeper
point, however, is that between them they articulated an alternative view of
mental illness, antagonistic to the traditional disease model and calling for new
ways of looking at the disordered person. In doing so, each in his own way
opened up for discussion a topic that continues to be a central issue for
psychiatry and which—though Laing would have rejected the jargon—can be
dryly stated in the form: Should psychological disorders be construed
dimensionally or discontinuously?
Tracing the more recent history of that debate in psychiatry is instructive in
providing some insights into the subsequently diverging influences of Laing
and Eysenck. A landmark event was the publication of the DSM-III (American
Psychiatric Association, 1980). A thorough revision of the previous, DSM-II,
version of the manual (American Psychiatric Association, 1968), it reflected, in
nosology, the views on etiology of a "new wave," organic psychiatry; this had
now become more confident in talking about mental illnesses as "types of
disease," encouraged by advances in neuroscience and related disciplines that
seemed to promise answers lacking in earlier times. Nonetheless, from its
introduction the DSM-III had its critics, among them both Laing and Eysenck.
The former's criticism, though later published (Laing, 1986), was first
delivered, as one of his biographers, Clay (1996), notes, in a drunken speech
in Glasgow, obscuring the more sensible ideas that Laing was attempting to
convey; for example his charge, which we can now see to have been valid, that
the advent of DSM-III signaled increasing, and for some unacceptable,
medicalization of mental suffering. But even if Laing had presented his views
more coherently he is unlikely to have made much impression. For, although it
is probably an exaggeration to claim a backlash against Laing's radical ideology
366 Further Eysenckian interests

as a primary motivation for the resuscitated organic psychiatry, it is the case

that he and his followers were by then drawing to themselves either
indifference or scathing comment: they were simply out of time.
Eysenck's criticism—incorporated into a seminal review of reactions to
DSM-III (Eysenck, Wakefield, and Friedman, 1983)—was, predictably, more
logical, incisive, and empirically grounded. He and his co-authors examined
regular questions like reliability and validity, as well as the theoretical
assumptions behind the new glossary. On the last point, they noted that most
adverse comment had focused on the fact that, while intended to be
atheoretical, the DSM-III actually relied heavily " ... on the 'medical model'
or categorical approach to mental disorders." Responding to this in then-
suggestions for further revisions to the manual, Eysenck et al. included a strong
recommendation that categorical diagnosis should be replaced in future by
dimensional assessment. Although this has not yet come about, it is noteworthy
that the DSM-IV task force did recognize the importance of the question, at
several points (American Psychiatric Association, 1994). Thus, in the Intro-
duction to the DSM-IV the limitations of the categorical model are openly
discussed and the dimensional approach acknowledged as a serious alternative:
Numerical dimensional descriptions are much less familiar and vivid than are the
categorical names for mental disorders. Moreover there is as yet no agreement on the
choice of the optimal dimensions to be used for classification purposes. Nonetheless,
it is possible that the increasing research on, and familiarity with, dimensional
systems may eventually result in their greater acceptance both as a method of
conveying clinical information and as a research tool. (p. xxii)
Later, in the coverage of the personality disorders, the matter is discussed
again:
The diagnostic approach used in this manual represents the categorical perspective
that Personality Disorders represent qualitatively distinct clinical syndromes. An
alternative to the categorical approach is the dimensional perspective that Personality
Disorders represent maladaptive variants of personality traits that merge impercept-
ibly into normality and into one another. There have been many different attempts to
identify the most fundamental dimensions that underlie the entire domain of normal
and pathological personality functioning ... The relationship of the various dimen-
sional models to the Personality Disorder(s) diagnostic categories and to various
aspects of personality dysfunction remains under active investigation, (pp. 633-634)
The conservatism of the above remarks has presumably caused some
eyebrows to be raised among psychologists working in the field; and the
observation about the "optimal dimensions" not yet having been found surely
brought a wry smile to Hans Eysenck's lips! Nevertheless, it is satisfying that
the dimensional model of mental illness he helped to pioneer is now beginning
to have at least some tangible influence on psychiatric thinking.
 Eysenck's contribution to understanding psychopathology 367

With respect to Eysenck's particular part in that process, one reason is

undoubtedly his sheer persistence and single- (and bloody-!) mindedness. For
all of his arguments against the medical model—mustered again to level criti-
cism at DSM-III—had been well rehearsed by him decades before, notably in
his own chapter in the Handbook of Abnormal Psychology (Eysenck, 1960b). But
the major reason—and a clear cause of R. D. Laing's declining influence—was
Eysenck's thoroughly biological stance on the etiology of mental illness; it is a
perspective that for some years now has chimed well with that of mainstream
psychiatry. Admittedly, when examined closely there are still some crucial
differences between the medical psychiatric approach to the biology and that of
what generically might now be called the "Eysenckian school." But the match is
close enough and the scope of research in the area sufficiently generous to
allow—for the moment at least—a range of models to exist side by side.
Dimensional description of individual variation and a biological explanation
of the underlying dynamics, between them, therefore capture the essence of
Eysenck's account of human nature, normal and abnormal. As in other areas of
his research activity, this twin-faceted theory provides the key to evaluating his
contribution to our understanding of psychopathology. It also helps to define
the structure of the rest of this chapter. Logically it seemed best to arrange the
material in a roughly temporal order, starting with Eysenck's concentration,
first, on the "less serious" disorders, then moving on to his relatively later work
on more severe, psychotic, illness. This will also make it easier to evaluate
Eysenck's ideas against others that have emerged over the same time span and
so work towards a view of where the strengths and weaknesses of his current
theory lie.
One other point should be made at the outset. Precisely because of the close
connection between normal and abnormal proposed by Eysenck, much that is
relevant to our account has already been presented, in other guises, by fellow
contributors to this volume. The discussion here will therefore assume some
knowledge of that other material and mostly concentrate on features of
Eysenck's theory that relate specifically to the abnormal domain.

2. THE TWO-DIMENSIONAL THEORY

2.1 Disorders of E and N
Although Eysenck's theorizing has always addressed the full range of mental
illness, the bulk of his early work was concerned with the less serious disorders.
The precise meaning of that term as currently usable within the Eysenckian
framework will be defined more precisely below. For the moment suffice to say
that the theory originated in Eysenck's attempt to extend into psychopathology
work on the two normal personality dimensions of introversion-extraversion
(I-E) and neuroticism (N). According to the first statements of theory
(Eysenck, 1947), the pathological counterparts of these dimensions were
368 Further Eysenckian interests

considered to be, on the one hand, "dysthymia," representing High N/Low E,

and, on the other, "hysteria," representing High N/High E. Discussion of the
model in that form is complicated, however, by several subsequent develop-
ments. I am referring to shifts of emphasis within the theory itself and to
general changes over the years in the meaning and/or usage of the terms
"dysthymia" and "hysteria."
In current psychiatric practice, and as listed in DSM-IV, "dysthymic
disorder" refers specifically to a state of depression; one that is long-standing,
but not severe enough to be considered major depressive illness. The
Eysenckian usage has always been much wider. It certainly includes mild—
or what in some older classifications would have been described as "reactive"
or "neurotic"—depression; but it also covers all of the anxiety-based neuroses,
such as phobias and generalized anxiety, as well as obsessive-compulsive
disorder. In personality terms, Eysenck was right to argue for lumping all of
these conditions together under one broad category of "dysthymia" in his
sense. The evidence consistently shows them all to fall clearly in the High N/
Low E quadrant of his two-dimensional system, as assessed by questionnaire
(Eysenck, 1959; H. J. Eysenck & S. B. G. Eysenck, 1965; Claridge, 1967); this
includes, despite some early disagreement on the matter, obsessional neurosis
[see Slade (1974) for a discussion of this point]. From an etiological perspective
the grouping into a single descriptive category has also tended to encourage
the idea that a unitary, or at very least overlapping, set of causal processes
might contribute to all High N/Low E disorders.
The High N/High E equivalent has always been more problematical. For one
thing, Eysenck's initial association of this combination of traits with hysteria, as
the major criterion (defining) group of disorders, turned out to be an
ambiguous choice. Certainly it was understandable historically, in following a
traditional dichotomy between psychasthenic (dysthymic) and hysterical forms
of neurosis (Jung, 1923), but "hysteria" has always been a difficult concept,
covering a wide range of psychologically disordered states (Merskey, 1995;
Roy, 1982). These have included the mimicking of medical conditions—for
example, the so-called "conversion" reactions—as well as syndromes closer to
descriptions of the "acting-out," hysterical personality. This heterogeneity was
reflected in the results of early questionnaire studies, which demonstrated that
levels of extraversion and neuroticism among patients with the hysteria
diagnosis often did not conform to Eysenck's prediction; scores were
frequently lower than expected on one or both scales (Sigal, Star, & Franks,
1958; McGuire, Mowbray, & Vallance, 1963; Claridge, 1967).
"Hysteria" of course no longer forms part of the official psychiatric
nomenclature, presumably having been dropped because of its offensive
overtones (especially regarding women). Although the substituted arrange-
ment for the same conditions does not fully solve the diagnostic confusion in
psychiatry—for example, the DSM listing of conversion reactions under
 Eysenck's contribution to understanding psychopathology 369

somatoform, rather than dissociative, disorders seems odd—it does help to

resolve matters with regard to Eysenck's theory. "Hysteria" is now spread
across both DSM Axes I and II, the former covering the classic conversion and
medically related neurotic syndromes; these are illnesses that appear to fall
outside the scope of an Eysenckian analysis, which is currently unable to offer a
good explanation of their underlying mechanisms. More relevant is the other,
Axis II, part of the previous "hysteria" designation, represented as the
"histrionic" form of personality disorder; the near equivalent of this in the
earlier literature was "hysterical personality" which, unlike other forms of
hysteria, does fit more clearly into Eysenck's dimensional description, as High
N/High E (Ingham & Robinson, 1964).
In the DSM classification histrionic personality disorder forms part of
Cluster B—the so-called "erratic-dramatic" group—and there are good
empirical and theoretical reasons (Farmer & Nelson-Gray, 1995) for believing
that the High N/High E quadrant also encloses some or all of other, comorbid,
personality disorders: narcissistic, borderline, and antisocial. Of course, the last
of these—under the "psychopathy" label—has for some time been regarded in
Eysenckian theory as the main criterion group for neurotic extraversion. Its
association to other Cluster B disorders simply serves to broaden the scope of
the theory's ability to deal with a range of personality deviations.
Here it is necessary to remind the reader of one point, however. Discussions
of psychopathy within Eysenckian theory have been complicated in recent
years by reference of them to the third, Psychoticism (P), dimension—implying
a connection to psychotic disorder. This will be ignored for the moment, in
order to evaluate the E/N model, applied to the two major classes of "less
serious" disorders, as defined here; viz. the dysthymic, mostly anxiety-based,
neurotic conditions and what, for want of a better descriptor, I will call the
hysterico-psychopathic forms of personality disorder.

2.2 Biological explanations

If one needed to pinpoint a single sign of Eysenck's scientific genius it would
lie, I believe, in his early conjoining of two perspectives on individual
differences: one stemming from the Western psychometric tradition of
descriptive personality measurement, the other to be found in the Pavlovian
theory of nervous types (Eysenck, 1955, 1957). The extension to the abnormal
domain naturally followed from the clinical applications already existing in
both streams of thought.
At this temporal distance, Eysenck's use of the Pavlovian model to devise his
own biological account of personality might seem quaint, given the peculiar
terminology in which the Russian theory was couched. His "translation" was
certainly cavalier, reducing to a simple statement about E and excitation-
inhibition what was subsequently revealed as a highly intricate theory (Gray,
370 Further Eysenckian interests

1964). But Eysenck's grand gesture of simplification was actually another sign
of his brilliance. For, although formulations of individual differences based on
"proper" nervous-type theory staggered on for a while (Nebylitsyn & Gray,
1972; Mangan, 1982), they never really caught on in the West, and in any case
have had little lasting impact in abnormal psychology.
The advantage of Eysenck's slimmed down version of the Russian theory
was twofold. First, it led to straightforwardly testable hypotheses about
putative nervous system differences in relation to personality—and to an
accumulation of a vast array of information (Eysenck, 1960c, 1967; and
reviewed elsewhere in this book). Secondly, when the time came—as it in-
evitably did—to re-jig the theory as a more Western-style arousal/activation
model the translation was, literally, little more than a linguistic exercise. In a
sense it scarcely mattered what the terminology was: in either form, the theory
was able to predict, among appropriately selected patient samples at least,
quite major differences in biological status, on a wide range of psychophysio-
logical and other experimental variables (Eysenck, 1967; Claridge, 1967).
In pure theory terms, a more significant question was—and still is—how far
our understanding of these clinical disorders was actually facilitated by
Eysenck's conceptualization of them as extreme forms of introversion and
extraversion. The logic of testing Eysenck's two-dimensional theory is that, on
laboratory measures, normal introverts and extraverts should behave in an
identical fashion to their patient—respectively dysthymic and hysterico-
psychopathic—counterparts. Yet many examples can be quoted in which this
is not so. A case in point is the measurement of barbiturate tolerance, or
"sedation threshold," a powerful experimental tool once widely used for
assessing individual differences in biological status. Studies of patient samples
strongly and consistently supported Eysenck predictions for his two criterion
groups: very high barbiturate tolerance (arousability) in dysthymics (all
subtypes) and very low tolerance in hysterico-psychopaths (Shagass & Jones,
1958; Claridge & Herrington, 1960; Claridge, 1967). Yet these patient
differences did not correspond at all to the results found in normal subjects,
where no simple relationship between sedation threshold and introversion-
extraversion (or indeed neuroticism) was ever found.
Such mismatches between the normal personality theory and the clinical
findings were partly explained by the interaction of introversion-extraversion
with neuroticism. This was revealed by doing what Eysenck (1967) called "zone
analysis," in which data are considered separately according to different
combinations of E and N. In the case of the sedation threshold measure
referred to above, the method was informative in showing just such an
interaction between E and N. It turned out that, among normal subjects,
sedative drug tolerance is related to extraversion in opposite directions,
depending on the level of neuroticism (Rodnight & Gooch, 1963; Claridge &
Ross, 1973; Claridge, Donald, & Birchall, 1981). Notably, the fit to Eysenck's
 Eysenck's contribution to understanding psychopathology 371

theory is among High N individuals, where introverts show greater tolerance

than extraverts, consistent with the patient findings. A typical observation, this
kind of interaction clarifies some of the uncertainty in Eysenck's theoretical
account; but it still leaves some worrying questions.
For one thing, explaining the biology of such relationships is still problem-
atical if one follows Eysenck's own (1967 and still unrevised) theory: that
separate causal processes of ARAS "arousal" and limbic "activation" form the
underlying determinants of E and N. Precisely how these two systems interact
to produce the observed psychopathology—or its experimental correlates—is
hard to discern. This in turn raises the difficulty of knowing precisely what
neuroticism is. Certainly it is well established statistically as an orthogonal
descriptive factor. And it is a convincing—almost "natural"—clinical construct
when applied to the dysthymic (High N/Low E) neuroses. But it is not easy,
within Eysenck's formulation of N, to visualize what elements these anxiety-
based, high arousal, states share with their biologically quite opposite
counterparts, the so-called "extraverted" disorders. In those cases,
"neuroticism" seems to be associated with, if anything, deficiencies in
arousability, leading to maladaptive "sensation seeking" behaviors of the
kind brought to our attention by Zuckerman (1979).
Here, for some critics, a better solution to such problems was achieved in
Gray's revision of Eysenck (Gray, 1981). The fusion of I and N to produce
"Anxiety" and E and N to produce "Impulsivity" does seem to make sense, for
two reasons. First, it utilizes two personality constructs which, it could be
argued, connect more manifestly than E and N to the associated psycho-
pathology: the transition from normal to abnormal is easier to conceptualize.
Second, the proposed new dimensions seem to map better onto possible
underlying brain systems, at least in the case of anxiety (Gray, 1982).
Although presumably not intended to do so, the Gray modification also
highlights an evident feature of this class of theory, including Eysenck's own.
As discussed in more detail elsewhere (Claridge, 1994), current biological
formulations of personality—at least those within the Eysenckian school—are
probably better construed as theories of temperament (see also chapter 4), viz.
as referring to relatively low-level features of individual variation. "Low level"
is used here in three senses; (1) unelaborated in the hierarchical structure of
adult personality, referring to relatively simple trait clusters; (2) reflecting
phylogenetically older brain systems, especially interpreted for their role in
determining emotional and motivational differences; (3) ontogenetically
primitive, corresponding to the early temperamental features studied in
developmental psychology (Buss & Plomin, 1984).
Interpreted in this way, there are clear implications for what we might expect
such theories to tell us about psychopathology. At a structural and descriptive
level one kind of knowledge they are likely to be able to provide concerns
predisposition to different forms of disorder: for example, the extent to which
372 Further Eysenckian interests

an individual's phobic state stems from an anxious temperament or

psychopathic behavior is explained by inherent impulsivity (Claridge, 1995).
Here, from a purely measurement point of view the information drawn upon is
not far removed from questionnaire scores, except in being more objective
and—if of a biological nature—closer to the putative genetic or neuro-
developmental substrate of the disposition in question.
But the theories under consideration claim more than that. They also intend
to establish the "dynamics"—viz. the symptom mechanisms—of psychological
disorder.

2.3 Dynamics of disorder

For Eysenck, like most of his followers, "dynamics" of psychological disorder
has always referred strictly to the behaviorist model, with symptoms being seen
as the result of dysfunctional processes of learning. The main critic of
Eysenck's version of this idea—strength of conditionability as the principal
mediator of psychopathological differences—has been Gray (1981; and
chapter 3). Gray's alternative—differences in reinforcement sensitivity, rather
than conditionability per se—is persuasive, both on empirical grounds and
because it seems better able to deal with the full range of disorders associated
with the E and N dimensions (though more especially those that are anxiety
based). In neither case, however, is there much scope for considering psycho-
logical (cognitive) factors, in the complete self-report sense that has come to
shape a good deal of thinking in clinical psychology. This therefore is an
important comparison that we need to make in judging the contribution of the
Eysenckian approach.
A first point to emphasize is that the outcome of such comparisons is bound
to depend on the kind—or even individual instance—of disorder under
consideration. In some cases the behaviorist model seems perfectly satis-
factory; in others woefully inadequate. A good example of the former is simple
phobias, where biological/behaviorist accounts contain many constructs that
still prove helpful in understanding etiology and in formulating treatment
methods. [Here we should not forget Eysenck's early contributions that gave
stimulus to the establishment of the behavior therapy movement (Eysenck,
1960a; see also the evaluation of this by Rachman (1981), and the debate of it
between Barbrack and Franks (1986) and Lazarus (1986).] It is even true that
some ideas from more cognitive (e.g., attentional bias) formulations of anxiety
(Williams, Watts, MacLeod, & Matthews, 1988) can be incorporated into
behaviorist models of phobic (and obsessional) behavior (Gray, 1982).
Yet the dividing line between biological/behaviorist and cognitive accounts
of disorder remains well defined; and the problem for the former is that the
majority of psychological disorders seen in clinical practice fall beyond the
scope of even the more sophisticated versions of historical behaviorism. Most
 Eysenck's contribution to understanding psychopathology 373

conditions are complexly determined and truly psychobiological—or even

biosocial—in nature; they are therefore not generally open to any single
explanation (Claridge, 1994). Usually, in analyzing their dynamics, it is
necessary to constantly switch models and terminology, depending on what
aspect is of concern or interest; only in that way is it possible to get a full
appreciation of all of the possible influences in etiology or the likely best focus
for treatment. An example—that of mild depression—helps to make the point.
The theoretical case can be illustrated as a series of increasingly "reduc-
tionist" questions, in the form: Why doesn't everyone exposed to, or
characterized by, X get depressed? Broad social factors, such as unemploy-
ment or poor housing, would start the sequence and more individual life events
continue it (Brown & Harris, 1978). At this stage some relevant personality
factors could be quoted as the reason why one individual reacts with clinical
disorder and another is unaffected. A likely personality dimension to be chosen
is neuroticism, which correlates strongly with the predisposition to and clinical
outcome in depression (Duggan, Lee, & Murray, 1990; Williams, 1992).
Psychologically (i.e., prior to any need for a biological analysis), neuroticism
can be construed as a vulnerability marker for depression that reflects a
tendency to adopt a negative cognitive set towards events and the self (Beck,
1976); this possibly arises from early experiences of failure and/or criticism
(Martin, 1985), or in some cases serious abuse (Mullen, Romans-Clarkson,
Walton, Herbison, 1988). But, again, some shrug off such childhood trauma;
only those temperamentally primed to overreact—those highly "sensitive to
punishment" (Gray, 1982)—will perhaps be vulnerable: first, to anxiety and
then, as a consequent end-state, to depression. This enhanced biological
susceptibility is undoubtedly subject to genetic influences (Loehlin, 1992).
Judged against this somewhat idealized synthesis of the factors affecting
depression, it is easy to see where the contribution of Eysenckian theory to our
understanding of psychopathology lies. It consists in helping to describe and
explicate a certain set of features, lying more towards the biological end of
what in most disorders is a chain of etiological events. This apparent emphasis
on the limitations of Eysenck's theory and its derivatives is not a criticism, but
rather an observation: that there probably cannot ever be any comprehensive
account of psychopathology, couched in the terminology of a single theory.

2.4 Conclusions
As those who have followed its progression over the years will know (and as
intimated here), the "strong" part of Eysenck's two-dimensional theory, when
applied to psychopathology, is that concerned with the dysthymic disorders. It
is no coincidence that above I chose depression to explore Eysenck's
contribution to the field. Admittedly, a not dissimilar synthesis could be
attempted for one or more of the High N/High E personality disorders, such as
374 Further Eysenckian interests

psychopathy or hysterical personality, since comparably varied accounts of

them have also been offered; including, recently, cognitive explanations
developed out of already established psychological approaches to depression
(Beck & Freeman, 1990). But generally speaking the understanding of these
conditions has lagged behind the advances made with respect to the anxiety-
based neuroses. This fact was mirrored in Eysenck's theory—until, that is, work
began in earnest on his third dimension.

3. INTO THREE DIMENSIONS

3.1 Introduction
The attempt to explain certain forms of personality disorder—and the
limitations this reveals in Eysenck's two-dimensional framework—not only
acts as a bridge between the previous section and the second half of this
chapter; it also identifies a major theme in Eysenck's handling of the more
severe, psychotic, disorders. His view that there is a common primary etiology
to antisocial behavior and psychosis is unusual and deserves careful scrutiny
against alternative formulations. First, however, it is necessary to consider a
prior and quite separate contribution that Eysenck made to the study of serious
mental illness—concerning its nosology.

3.2 One psychosis or several?

Eysenck's (1950,1952) early excursion into research on psychotic disorder was
typically controversial, in two respects: it pursued a dimensional perspective
and it rejected the contemporaneous medical distinction between schizo-
phrenia and manic depression, in favor of the notion of Einheitpsychose (or
unitary psychosis). This idea, that there is just a single form of insanity, was
actually a popular view in the nineteenth century (Berrios, 1995); but, by the
time Eysenck began writing on the topic, psychiatric classification had long
been in the grip of the Kraepelinian binary model.
Eysenck's (1952) own experimental and factor analytic approach to the
question was not ideal. Although it neatly demonstrated the technique of
"criterion analysis," the laboratory measurements entered into the factor
analysis were atheoretic and more indices of general intellectual competence
than pointers to anything of genuine etiological interest. Nevertheless, the
study achieved its two objectives, in showing: (1) continuity between the
normal and the clinically psychotic and (2) the existence of a single, unipolar
dimension of psychoticism, as against the bipolar schizothymia-cyclothymia
continuum postulated by Kretschmer (1925). At the time, and for many years
afterwards, these conclusions were ignored; or, in psychiatry, Eysenck's
 Eysenck's contribution to understanding psychopathology 375

position was misrepresented by commentators focusing on the more

contentious, dimensional aspect of his theorizing, rather than on what he
had to say about the binary/unitary issue (e.g., Kendell, 1974).
In recent years the Kraepelinian model has attracted increasing criticism
from psychiatrists themselves, some of whom have considered the unitary
explanation of psychotic disorder to be a possibly more viable alternative
[Kendell, 1991; Crow, 1986; see also Taylor (1992) for a useful review]. The
contemporary arguments are certainly persuasive, stemming from a variety of
data about the lack of distinctiveness of schizophrenia, when compared with
bipolar affective (manic depressive) psychosis. This includes the failure to find
a clear point of rarity in clinical symptomatology (Kendell & Brockington,
1980), and the presence therefore of the mixed, schizoaffective syndrome; the
interchangeability of treatments (Klein & Fink, 1963; Overall, Hollister, &
Meyer, 1964; Delva & Letemendia, 1982; Abraham & Kulhara, 1987); and in
genetics, despite some cautioning against outright rejection of the binary
model (Stromgren, 1994; LaPierre, 1994), evidence for intermingling of
familial liability to the two forms of psychosis (Baron & Gruen, 1991). To
which Kendell (1991) has added the important observation that, among the
many biological abnormalities claimed for schizophrenia over the years, none
has proved specific to that disorder, but can also be found in a proportion of
patients with affective illness. Interestingly, this "rediscovery" by psychiatrists
of Einheitpsychose theory has proceeded with virtually no acknowledgment of
the fact that Eysenck anticipated its revival by half a century!
Although some psychiatrists are now more ready to embrace the idea of
overlap, or even continuity, between the clinical syndromes, few, if any, seem
able to cope with a fully dimensional view of psychosis, in the sense of
psychotic features stretching back into the normal personality. Indeed, as
discussed later, this has even proved problematical for some psychologists
working in the area. In the experimental field it is a perspective really only to
be found among those pursuing Eysenck's theory directly, or those whose
thinking has been shaped by it. The "modern" phase of such research is
represented in work that originated in the development of questionnaire
instruments (the P-scales) for measuring psychoticism descriptively (H. J.
Eysenck & S. B. G. Eysenck, 1976; see also chapter 6). Some consideration of
these instruments is therefore central to our understanding of Hans (and, with
him, Sybil) Eysenck's more recent contribution to psychosis research.

3.3 The significance of P

From the outset, worries were expressed about the construct, criterion, and
face validity of the P-scales (see Claridge, 1981; 1983 for earlier reviews). The
issue was articulated at the time as a fairly straightforward question: Do the
cold, heartless, aggressive, personality traits measured by the later—EPQ (and
376 Further Eysenctdan interests

subsequent EPQ-R)—versions of the scales really capture an important

feature of psychotic disorder (or any of its forms)? One opinion (e.g.,
Zuckerman's, 1989) was that P is actually a dimension more related to
psychopathy. Accordingly, its inclusion in Eysenck's theory was considered a
significant advance because it powerfully extended the explanation of the
Cluster B personality disorders, especially the antisocial form; but, it was
argued, it is misleading to label the dimension "psychoticism" because its
descriptive features, as represented in the P-scale, have little to do with
psychosis. An alternative view, including of course Eysenck's own, acknowl-
edged the strong association of psychopathy to P, but went on to argue that it is
precisely because of this connection that P can be regarded as a good measure
of psychotic traits; for, it is claimed, psychotic and antisocial behavior are
themselves closely interlinked, genetically and in other biological and
behavioral respects.
Eysenck (1992) is now quite explicit on this point, postulating a continuum
running from altruistic, socialized, empathic attitudes at one extreme, through
aggressive criminality and psychopathy, to affective, schizoaffective, and
schizophrenic disorder, at the other (see Figure 6.2 in chapter 6). This is a
bold theoretical position and some parts of it certainly find support in
Eysenck's review of the evidence. Schizoid traits—very relevant to
schizophrenia—are definitely "P-like" in quality; there are undoubted familial
links between psychosis and psychopathy; and, with respect to the personality
disorders in general, both the borderline (BPD) and schizotypal (SPD) forms
stand as a clear indication that psychiatric conditions constitute a spectrum
which spans both the psychotic and the nonpsychotic—or, in DSM terms, both
Axis I and Axis II. Looked at in this way, the Eysencks' choice of how to
characterize their third dimension seems reasonable, especially given their
intention to do so from a personality, rather than a clinical, point of view.
Still, some worries remain about the P-scale—and therefore about the way
"psychoticism" is conceptualized within Eysenck's theory. Specifically, is it
really being argued that aggressivity—fundamental to the definition of P—is a
sufficient and/or necessary cause of psychotic illness? Or, put another way, how
much of the variance of risk for psychosis can be said to be due to such traits?
One set of evidence relevant here comes from inquiries into the supposed
association between schizophrenia and violent behavior. Certainly, there exists
a commonly held perception of the schizophrenic as highly unpredictable
(Levey & Howells, 1995), if not downright dangerous (Furnham & Rees,
1988), a stereotype powerfully reinforced by occasional headline-grabbing
stories of berserk killers. But how far does this conform to the reality about the
majority of schizophrenic individuals?
One difficulty with assessing such evidence is that in many studies the
apparent association found between crime, including violent crime, and mental
illness disappears once factors such as age, sex, race, and social class are
 Eysenck's contribution to understanding psychopathology 377

controlled for (Monahan & Steadman, 1983). Furthermore, sampling bias can
be introduced into studies of hospitalized patients, since violence itself is often
used as a criterion for committal under mental health legislation (Monahan,
1992). Even so, there is work which, taking account of these influences, does
suggest some genuine statistical association between schizophrenia and
aggressive behavior (Swanson, 1994). More problematical for a primary etio-
logical theory of psychosis—such as that proposed by Eysenck—are (1) the
extent of this association and (2) its interpretation.
Regarding the first question, most studies of relatively unselected hospital or
community samples indicate that, even if schizophrenics do commit more
aggressive acts, most of these tend to be of a minor nature or, if more serious,
in absolute terms rather rare (Buckley, Walshe, Colohan, & O'Callaghan,
1990; Modestin & Ammann, 1996). For example, the latter authors reported a
highly significant difference in the incidence of violent crime among schizo-
phrenics compared with matched controls. But inspection of their data reveals
that this amounted to actual rates of, respectively, 15 and 3 in samples of nearly
300 subjects: actuarially, but scarcely clinically, significant!
The most important findings in this literature concern the causes of violence
among the minority of schizophrenics who are seriously aggressive. Here
several facts emerge. First, there is a considerable consensus that the violence
is frequently secondary to, or motivated by, delusional beliefs of persecution or
distorted perception of the behavior of others, often a significant person in the
individual's life (Humphreys, Johnstone, MacMillan, & Taylor, 1992; Nestor,
Haycock, Doiron, J. Kelly, & D. Kelly, 1995; Junginger, 1996). Second, some of
the excess violence among schizophrenics reflects the presence of clinically
comorbid, but not necessarily etiologically shared, psychopathic disorder
(Rasmussen, Levander, & Sletvold, 1995). Indeed, in one instance where this
was deliberately examined, schizophrenia alone was associated with a lessened
risk of violence (Rice & Harris, 1995). Third, aggressive acts are often trig-
gered in schizophrenics (as they are in anyone else) by alcoholism or other
forms of substance abuse and may therefore be only indirectly attributable to
the psychotic state itself (Rice & Harris; Rasmussen et al.; Nestor et al.). In the
same category, though more specific to schizophrenia, are the unwanted
behavioral consequences of neuroleptic medication. Especially pertinent here
is the relatively common side-effect of akathisia, with its subjective experience
of internal turmoil, restlessness, irritability, and impulsivity (Casey, 1994; Awad
& Hogan, 1994). Akathisia has frequently been considered responsible for
both impulsive suicidal acts and violent aggressive behavior among schizo-
phrenics (Barnes, 1990).
Taken together, the effects described can probably account for most of the
relatively low degree of violence found in schizophrenia and it seems unlikely
that aggressiveness per se constitutes a primary risk factor for psychotic
disorder. By extrapolation, therefore, the traits associated with psychoticism in
378 Further Eysenckian interests

Eysenck's theory cannot be considered as uniquely, or importantly, "psychotic"

as is claimed. On the other hand, for reasons already discussed, the P
dimension clearly is relevant to our understanding of serious mental illness.
This almost certainly comes about through the influence P as one of a number
of factors that interact to define the risk for psychotic breakdown. Here
alternative approaches to the dimensionality of psychosis are informative.

4. PSYCHOTICISM AND SCHIZOTYPY COMPARED

In parallel with, but historically independent of Eysenck's development of

psychoticism as a personality dimension, work has proceeded from a more
clinical viewpoint on the concept of "schizotypy" (Meehl, 1962). Although
superficially similar, the two constructs have traditionally differed somewhat.
Most obvious has been the concentration in schizotypy research on the specific
question of risk for schizophrenia, rather than psychosis as a whole (Meehl,
1990). More subtly, and perhaps reflecting its North American origins, the
dimensionality in schizophrenia has also been mostly interpreted within a
framework of spectrum illness, rather than in the context of healthy individual
differences. That is to say, schizotypy—even when accepted as occurring in
"normal" people—has been construed as a forme fruste of schizophrenic
disease (continuous with schizotypal personality disorder), not as a truly
dimensional personality characteristic. This theoretical stance has dictated a
dichotomous, taxonometric approach to measurement (Lenzenweger &
Korfine, 1995) and, compared with European research, a more medically
orientated methodology for investigating the experimental correlates and
biological underpinnings of schizotypy (Lenzenweger, 1994).
In recent years, through an active exchange of ideas and research findings,
some convergence has occurred between the two schools of thought, with
benefits to both (Raine, Lencz, & Mednick, 1995). Many more questionnaire
scales have become available, especially those developed, or inspired by, the
work of the Chapmans in Wisconsin (Edell, 1995); they, interestingly, have also
increasingly preferred the broader term "psychosis proneness"—a near
equivalent of "psychoticism"—rather than "schizotypy," to designate their
scales. In return, those of us hi Europe influenced by Eysenckian theory have
been prompted to criticize the quasi-dimensional view of psychosis risk
adopted by American colleagues and sought to bring schizotypy more into the
domain of normal personality (Claridge, 1997).
A particular consequence of this rapprochement has been a brisk trade in
questionnaire data, including many attempts to establish the structure of
schizotypy through factor analysis of scales which vary considerably in item
content (see Mason, Claridge, & Williams, 1997, for a recent review).
Sometimes the P scale has been included in such studies, enabling us to judge
 Eysenck's contribution to understanding psychopathology 379

where "psychoticism," as defined by Eysenck, lines up against other factors.

The most consistent finding has been the appearance in all analyses of a first,
strong, factor covering unusual perceptual and other cognitive experiences,
contiguous with the positive symptoms of schizophrenia. A comparable nega-
tive symptom component has also generally emerged, though the remaining
factor structure has naturally varied slightly depending on the number and kind
of scales included in the analysis. In the largest study to date, in terms of
sample size and scales (including the P scale) used, four factors were identified
(Claridge et al., 1996). One of these—labeled "asocial behavior"—was mostly
defined by P and was separate from other components, which loaded on more
clinically based scales. P therefore seems to form one element in a
multidimensional structure that describes, to use the Chapmans'
terminology, "psychosis proneness"; or we might label it "psychoticism," hi a
more generic sense than that adopted by Eysenck. Within that structure,
schizotypy—as a narrower construct—could be aligned with more specifically
schizophrenic aspects, as traditionally recognized in the Kraepelinian model.
It is fair to point out here that Eysenck has challenged the above
interpretation of the factor structure of "schizotypy." Conducting an analysis
of some appropriate questionnaire data, he reports three factors that map
directly on to his own E, N, and P dimensions, with "little trace of schizotypy
left, and no evidence of any single factor corresponding to such a factor"
(Eysenck & Barrett, 1993). But the latters' rejection of the schizotypy concept
is really too harsh, for three reasons. First, there is evidence that its
components appear even in the absence of an "Eysenckian" influence on the
data. Thus, in the Claridge et al. (1996) study referred to above the correlation
matrix—which had originally included all four EPQ scales—was reanalyzed,
retaining only the P scale. The same four-factor structure emerged, suggesting
that its previous form had not been strongly dictated by the presence of E and
N in the matrix. Second, as mentioned earlier, there is the ubiquitous
appearance in all analyses of psychosis proneness scales of a leading "positive
symptom" factor. Such a component is not phenomenologically represented in
the Eysenckian description of personality, including psychoticism, and it is
difficult to see how it could be derived from it. And, third, we cannot ignore the
fact that most of the experimental paradigms currently used to investigate
psychosis proneness focus on the positive symptom aspect: attempting to
explain the aberrant perceptions and beliefs that constitute its normal
equivalent has become a research priority. Indeed, it could be argued that it
is precisely this cognitive component—the susceptibility to weird experiences—
that forms the sought-after necessary condition of risk for psychosis.
Having said that, the component in question does not, by itself, appear to be
a sufficient condition. On the contrary, there is ample evidence that so-called
positive symptoms can exist in perfectly benign, even more than averagely
adaptive, form: as religious (Jackson, 1997) or other pleasurably felt
380 Further Eysenckian interests

experiences (McCreery, 1997); that fact prompted the latter author to coin the
term, "happy schizotype," to denote individuals who have all the "positive
symptom," but none of the pathological illness, features of schizotypy
(McCreery & Claridge, 1996). As suggested earlier, it is the interaction with
other variables—including other components of psychosis proneness—that is
probably crucial in deciding whether the outcome of risk is psychological
disorder, eccentricity, or simply healthy adjustment. This conclusion would be
supported by recent evidence from the Chapmans on the predictive validity of
their various scales (L. J. Chapman, J. P. Chapman, Kwapil, Eckblad, Zinser,
1994). As anticipated, from among their "positive symptom" scales Magical
Ideation did quite well as a predictor of psychosis at 10-year follow-up; but the
best prediction came when Magical Ideation was combined with their Social
Anhedonia scale. In other words, maximum psychotic breakdown depended on
the shaping of the primary cognitive disposition by the negative affect
associated with anhedonic personality traits. (It is worth noting here that the
latter might have some common variance with Eysenck's P; in their factor
analysis of schizotypy scales Raine & Albutt (1989) reported that Social
Anhedonia and P together defined a component of what they called
"anhedonic psychoticism.")

5. CONCLUSIONS

Looked at against the above background, the P construct begins to fall into
place. It might not have the status of a primary etiological factor in psychosis
that Eysenck attributes to it. On the other hand, in its capacity as a personality
dimension of strong (aggressive) affect it could be quite crucial in determining
actual psychotic breakdown or the appearance of behaviors that cause society
to label an individual schizophrenic. (Here we should recall the earlier
discussion about schizophrenic violence, the schizophrenia diagnosis, and the
role of co-existent delusional belief.) For a similar reason, Eysenck's
formulation of psychoticism also copes well with the borderline states; this is
especially true of antisocial personality disorder, which, more than full-blown
psychosis, lends itself to an explanation purely in terms of motivation,
temperament, and affect.
It is of course no coincidence that the limitations evident in Eysenck's three-
dimensional theory are similar to those discussed for the two-dimensional
version; that is to say, a relative disinterest in the more "psychological,"
introspectively accessed features of personality and its disorders. Instead, the
addition of psychoticism simply—and, one has to say, rather neatly—completes
a descriptive and causal framework of individual differences largely designed
around the biology of drive and emotion. The three elements of affiliativeness
(E), fearfulness (N), and aggression (P) together now provide, at a certain
 Eysenck's contribution to understanding psychopathology 381

level, a comprehensive account of human variation that is highly relevant to

our attempts to understand psychopathology. Where the theory is important is
in helping to define the parameters of disposition for different disorders,
rather than in leading us directly to an understanding of the symptomatology of
the individual case. But that is the intended style of the theory and how it
should be judged.

6. FUTURE DIRECTIONS

Construed in the way I have favored here—as a theory of temperamental

disposition—Eysenck's account is preeminent and surely provides the
definitive model within which certain obvious research developments will
proceed. One of these is the further exploration of genetic influences on
individual differences. The methodology of the by now anachronistically
named "new" genetics will almost certainly turn references to the heritability
of E, N, and P from statistical into biological statements: the inherited
component in a person's vulnerability to disorders with which the dimensions
turn out to be strongly associated will then be capable of being precisely
delineated. The future ethical issues to be faced as a result of such discoveries
are (fortunately or unfortunately) outside the scope of this discussion: they will
exist irrespective of the correctness or otherwise of Eysenck's theory. However,
one brief general observation is perhaps in order.
The negative (eugenics) consequences of genomics—the mapping of the
genome—are fairly familiar and constitute a danger to which society must be
constantly alert and must continually debate (Pelosi & David, 1989). But the
potential benefits should also be recorded. In a recent discussion of what they
perceive to be "the next psychiatric revolution," Farmer and Owen (1996)
rightly point out that establishing the genetics of a disorder can also elucidate
the exogenous influences in its etiology and treatment, by identifying features
of the environment to which a genotype is specifically sensitive. By way of
example, they quote family influences on the course of schizophrenia, as well
as other disorders which, despite being highly heritable, are responsive to
psychological manipulations. It will remain to be seen whether an increasingly
biological psychiatry proceeds along the well-meaning path that Farmer and
Owen try to map out for it; or whether it is content merely to continue seeking
organic causes and organic treatments for organic diseases. Here a neo-
Eysenckian theory could have a unique responsibility, in keeping alive a
genuine psychobiological approach to mental illness, in which the biological
and the psychological, the genetic and the environmental, and the objectively
observed and introspectively reported all have equal status in contributing to
our understanding. One important tradition in such types of theory—
grounding the explanation of disorder in a fully dimensional, personality-
382 Further Eysenckian interests

based model of etiology—makes them ideally fitted to that task. It would be a

pity—and an irony of history—if they were, instead, entrapped into pursuing a
narrower disease-based, more reductionist view, indistinguishable from that
currently being fostered in some quarters of psychiatry.
So, although Eysenckian theory will certainly continue to be more com-
fortable explaining things at the biological end, it will hopefully begin to find
ways of doing so that encompass data about mental illness not currently dealt
with very well. Mention has been made several times of the "cognitive"
dimension in psychopathology. This term has several different meanings,
however. It covers the mechanisms of persecutory delusions in psychosis; the
negative thought biases of depressed patients; the obsessive-compulsive's
ruminations; and the dissociative reactions of the severely traumatized. It
seems inevitable that these cognitive features of disorder will continue to be
examined at a purely psychological, experiential level. But a supplementary
biological account is also desirable, and perfectly feasible: in fact—by way of
illustration—all of the clinical phenomena just mentioned have been
incorporated into explanations which draw upon knowledge about lateralized
brain organization (Myslobodsky, 1983). That and other neuropsychological
perspectives on higher nervous system functioning have been relatively
neglected in Eysenckian theory; yet they are likely to prove essential to a
complete understanding of the biology of personality—and, by the same token,
psychopathology. Their integration with current models of the biology of
temperament seems a natural future direction for research in the area to take.

REFERENCES

Abraham, K. R. & Kulhara. P. (1987). The efficacy of electroconvulsive therapy in the

treatment of schizophrenia: a comparative study. British Journal of Psychiatry, 151,
152-155.
American Psychiatric Association. (1968). Diagnostic and statistical manual (2nd ed.).
Washington, DC: APA.
American Psychiatric Association. (1980). Diagnostic and statistical manual (3rd ed.).
Washington, DC: APA.
American Psychiatric Association (1994). Diagnostic and statistical manual (4th ed.).
Washington, DC: APA.
Awad, A. G. & Hogan, T. P. (1984). Subjective response to neuroleptics and the quality
of life: implications for treatment outcome. Acta Psychiatrica Scandinavica, 89
(Suppl. 380), 27-32.
Barbrack, R. & Franks, C. M. Contemporary behaviour therapy and the unique
contribution of H. J. Eysenck: Anachronistic or visionary? In S. Modgil & C. Modgil
(Eds.), Hans Eysenck. consensus and controversy (pp. 233-245 and 258-260). Lewes,
U.K.: The Palmer Press.
 Eysenck's contribution to understanding psychopathology 383

Barnes, T. R. E. (1990). Drug-induced akathisia and tardive dyskinesia. In M. Weller

(Ed.), International perspectives in schizophrenia. Chichester, U.K.: John Wiley.
Baron, M. & Gruen, R. S. (1991). Schizophrenia and affective disorder: Are they
genetically linked? British Journal of Psychiatry, 159, 267-270.
Beck, A. T. (1976). Cognitive therapy and the emotional disorders. New York:
International Universities Press.
Beck, A. T. & Freeman, A. (1990). Cognitive therapy of personality disorders. New York:
The Guilford Press.
Berrios, G. E. (1995). Conceptual problems in diagnosing schizophrenic disorders. In J.
A. Den Boer, H. G. M. Westenberg, & H. M. van Praag (Eds.), Advances in the
neurobiology of schizophrenia (pp. 7-25). Chichester, U.K.: Wiley.
Brown, G. & Harris, T. (1978). Social origins of depression, a study of psychiatric disorder
in women. London: Tavistock.
Buckley, P., Walshe, D., Colohan, H. A., & O'Callaghan, E. (1990). Violence and
schizophrenia: A study of the occurrence and clinical correlates of violence among
schizophrenic patients. Irish Journal of Psychological Medicine, 7, 102-108.
Burston, D. (1996). The wing of madness. The life and work ofR. D. Laing. Cambridge,
MA: Harvard University Press.
Buss, A. H. & Plomin, R. (1984). Temperament: Early developing personality traits.
Hillsdale, NJ: Lawrence Erlbaum.
Casey, D. E. (1994). Motor and mental aspects of acute extrapyramidal syndromes. Acta
Psychiatrica Scandinavica, 89 (Suppl. 380), 14-20.
Chapman, L. J., Chapman, J. P., Kwapil, T. R., Eckblad, M., & Zinser, M. C. (1994).
Putatively psychosis prone subjects 10 years later. Journal of Abnormal Psychology,
103, 171-183.
Claridge, G. (1967). Personality and arousal. Oxford: Pergamon Press.
Claridge, G. S. (1981). Psychoticism. In R. Lynn (Ed.), Dimensions of personality. Papers
in honour of H. J. Eysenck (pp. 79-109). Oxford: Pergamon Press.
Claridge, G. S. (1983). The Eysenck Psychoticism Scale. In J. N. Butcher & C. D.
Spielberger (Eds.), Advances in personality assessment (Vol. II) (pp. 71-114). New
Jersey: Erlbaum.
Claridge, G. (1990). Can a disease model of schizophrenia survive? In R. P. Bentall
(Ed.), Reconstructing schizophrenia (pp. 157-183). London: Routledge.
Claridge, G. (1994). Psychobiological models and issues. In S. Strack & M. Lorr (Eds.),
Differentiating normal and abnormal personality (pp. 137-157). New York: Springer.
Claridge, G. (1995) Origins of mental illness. Cambridge, MA: Malor Books.
Claridge, G. (Ed.) (1997). Schizotypy: Implications for illness and health. Oxford: Oxford
University Press.
Claridge, G. S. & Herrington, R. N. (1960). Sedation threshold, personality, and the
theory of neurosis. Journal of Mental Science, 106, 1568-1583.
Claridge. G. & Ross, E. (1973). Sedative drug tolerance in twins. In G. Claridge, S.
Canter & W. I. Hume (Eds.), Personality differences and biological variations. A study
of twins (pp. 115-131). Oxford: Pergamon Press.
Claridge, G., Donald, J. R., & Birchall, P. M. (1981). Drug tolerance and personality:
Some implications for Eysenck's theory. Personality and Individual Differences, 2,
153-166.
384 Further Eysencldan interests

Claridge, G., McCreery, C., Mason, O., Bentall, R., Boyle, G., Slade, P., & Popplewell,
D. (1996). The factor structure of "schizotypal traits": A large replication study.
British Journal of Clinical Psychology, 35, 103-115.
Clay, J. (1996). R D. Lang. A divided self. London: Hodder & Stoughton.
Crow, T. J. (1986). The continuum of psychosis and its implication for the structure of
the gene. British Journal of Psychiatry, 149, 419-429.
Delva, N. J. & Letemendia, F. J. J. (1982). Lithium treatment in schizophrenia and
schizoaffective disorders. British Journal of Psychiatry, 141, 387-400.
Duggan, C. F., Lee, A. S., & Murray, R. M. (1990). Does personality predict long-term
outcome in depression? British Journal of Psychiatry, 157, 19-24.
Edell, W. S. (1995). The psychometric measurement of schizotypy using the Wisconsin
scales of psychosis proneness. In G. A. Miller (Ed.), The behavioral high-risk paradigm
in psychopathology (pp. 3-46). New York: Springer.
Eysenck, H. J. (1947). Dimensions of personality. London: Routledge & Kegan Paul.
Eysenck, H. J. (1950). Schizothymia-cyclothymia as a dimension of personality I.
Historical review. Journal of Personality, 19, 123-152.
Eysenck, H. J. (1952). Schizothymia-cyclothymia as a dimension of personality II.
Experimental. Journal of Personality, 20, 345-384.
Eysenck, H. J. (1955). A dynamic theory of anxiety and hysteria. Journal of Mental
Science, 101, 28-51.
Eysenck, H. J. (1957). Dynamics of anxiety and hysteria. London: Routledge & Kegan
Paul.
Eysenck, H. J. (1959). Manual of the Maudsley Personality Inventory. London: London
University Press.
Eysenck, H. J. (Ed.) (1960a). Behaviour therapy and the neuroses. Oxford: Pergamon.
Eysenck, H. J. (1960b). Classification and the problem of diagnosis. In H. J. Eysenck
(Ed.), Handbook of abnormal psychology (1st ed.) (pp. 1-31). London: Pitman.
Eysenck, H. J. (Ed.) (1960c). Experiments in personality (Vols. I and II). London:
Routledge & Kegan Paul.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL.: Charles C.
Thomas.
Eysenck, H. J. (1992). The definition and measurement of psychoticism. Personality and
Individual Differences, 13, 757-785.
Eysenck, H. J. & Barrett, P. (1993). The nature of schizotypy. Psychological Reports, 73,
59-63.
Eysenck, H. J. & Eysenck, S. B. G. (1965). Manual of the Eysenck Personality Inventory.
London: Hodder & Stoughton.
Eysenck, H. J. and Eysenck, S. B. G. (1976). Psychoticism as a dimension of personality.
London: Hodder and Stoughton.
Eysenck, H. J., Wakefield, J. A., & Friedman, A. F. (1983). Diagnosis and clinical
assessment. Annual Review of Psychology, 34, 167-193.
Farmer, A. & Owen, M. J. (1996). Genomics: The next psychiatric revolution? British
Journal of Psychiatry, 169, 135-138.
Farmer, R. F. & Nelson-Gray, R. O. (1995). Anxiety, impulsivity, and the anxious-
fearful and erratic-dramatic personality disorders. Journal of Research in Personality,
29, 189-207.
 Eysenck's contribution to understanding psychopathology 385

Furnham, A. & Rees, J. (1988). Lay theories of schizophrenia. International Journal of

Social Psychiatry, 34, 212-220.
Gray, J. A. (1964). Pavlov's typology. Oxford: Pergamon Press.
Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.),
A model for personality, (pp. 246-276), Berlin: Springer.
Gray, J. A. (1982). The newopsychology of anxiety. Oxford: Clarendon Press.
Humphreys, M. S., Johnstone, E. C., MacMillan, J. F., & Taylor, P. J. (1992).
Dangerous behaviour preceding first admissions for schizophrenia. British Journal of
Psychiatry, 161, 501-505.
Ingham, J. G. & Robinson, J. O. (1964). Personality in the diagnosis of hysteria. British
Journal of Psychology, 55, 276-284.
Jackson, M. (1997). Benign schizotypy? The case of spiritual experience. In G. Claridge
(Ed.), Schizotypy: Implications for illness and health (pp. 227-250). Oxford: Oxford
University Press.
Jung, C. G. (1923). Psychological types (trans. H. G. Baynes). London: Kegan Paul.
Junginger, J. (1996). Psychosis and violence: the case for a content analysis of psychotic
experience. Schizophrenia Bulletin, 22, 91-103.
Kendell, R. E. (1974). The role of diagnosis in psychiatry. Oxford: Blackwell.
Kendell, R. E. (1991). The major functional psychoses: Are they independent entities or
part of a continuum? Philosophical and conceptual issues underlying the debate. In
A. Kerr & H. McClelland (Eds.), Concepts of mental disorder. A continuing debate
(pp. 1-16). London: Gaskell.
Kendell, R. E. & Brockington, I. F. (1980). The identification of disease entities and the
relationship between schizophrenic and affective psychoses. British Journal of
Psychiatry, 135, 243-248.
Klein, D. F. & Fink, M (1963). Multiple item factors as change measures in
psychopharmacology. Psychopharmacologia, 4, 43-52.
Kretschmer, E. (1925). Physique and character (trans. W. J. H. Sprott). London: Kegan,
Trench, and Trubner.
Laing, R. D. (1960). The divided self. London: Tavistock.
Laing, R. D. (1986). God and psychiatry. Times Literary Supplement, 23 May.
LaPierre, Y. D. (1994). Schizophrenia and manic-depression: separate illnesses or a
continuum? Canadian Journal of Psychiatry, 39, 559-564.
Lazarus, A. A. (1986). On sterile paradigms and the realities of clinical practice: critical
comments on Eysenck's contribution to behaviour therapy. In: S. Modgil & C.
Modgil (Eds), Hans Eysenck. consensus and controversy (pp. 247-257 and 260-261).
Lewes, U.K.: The Palmer Press.
Lenzenweger, M. F. (1994). Psychometric high-risk paradigm, perceptual aberrations,
and schizotypy: an update. Schizophrenia Bulletin, 20, 121-135.
Lenzenweger, M. F. & Korfine, L. (1995). Tracking the taxon: on the latent structure
and base rate of schizotypy. In A. Raine, T. Lencz, & S. A. Mednick (Eds.),
Schizotypal personality (pp. 135-167). Cambridge: Cambridge University Press.
Levey, S. & Howells, K. (1995). Dangerousness, unpredictability and the fear of people
with schizophrenia. Journal of Forensic Psychiatry, 6, 19-39.
Loehlin, J. C. (1992). Genes and environment in personality development. London: Sage.
Mangan, G. (1982). The biology of human conduct. Oxford: Pergamon Press.
386 Further Eysencfdan interests

Martin, M. (1985). Neuroticism as predisposition towards depression: A cognitive

mechanism. Personality and Individual Differences, 6, 353-365.
Mason, O., Claridge, G., & Williams, L. (1997). Questionnaire measurement. In G.
Claridge (Ed.), Schizotypy: Implications for illness and health (pp. 19-37). Oxford:
Oxford University Press.
McCreery, C. (1997). Hallucinations and arousability: pointers to a theory of psychosis.
In G. Claridge (Ed.), Schizotypy: Implications for illness and health (pp. 251-273).
Oxford: Oxford University Press.
McCreery, C. & Claridge, G. (1996). A study of hallucination in normal subjects I. Self-
report data. Personality and Individual Differences, 21, 739-747.
McGuire, R. J., Mowbray, R. M., & Vallance, R. C. (1963). The Maudsley Personality
Inventory used with psychiatric inpatients. British Journal of Psychology, 54,157-166.
Meehl, P. E. (1962). Schizotaxia, schizotypy, and schizophrenia. American Psychologist,
17, 827-833.
Meehl, P. E. (1990). Toward an integrated theory of schizotaxia, schizotypy, and
schizophrenia. Journal of Personality Disorders, 4, 1-99.
Merskey, H. (1995). The analysis of hysteria (2nd ed.). London: Gaskell.
Modestin, J. & Ammann, R. (1996). Mental disorder and criminality: Male schizo-
phrenia. Schizophrenia Bulletin, 22, 69-82.
Monahan, J. (1992). Mental disorder and violent behaviour: Perceptions and evidence.
American Psychologist, 47, 511-521.
Monahan, J. & Steadman, H. (1983). Crime and mental disorder: an epidemiological
approach. In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of
research (Vol. 4) (pp. 145-189). Chicago, IL: University of Chicago Press.
Mullen, P. E., Romans-Clarkson, S. E., Walton, V. A., & Herbison, G. P. (1988).
Impact of sexual and physical abuse on women's mental health. Lancet, 16 April,
841-845.
Myslobodsky, M. S. (Ed.) (1983). Hemisyndromes. Psychobiology, neurology, psychiatry.
New York: Academic Press.
Nebylitsyn, V. D. & Gray, J. A. (Eds.) (1972). Biological bases of individual behaviour.
New York: Academic Press.
Nestor, P. G., Haycock, J., Doiron, S., Kelly, J., & Kelly, D. (1995). Lethal violence and
psychosis: a clinical profile. Bulletin of the American Academy of Psychiatry and the
Law, 23, 331-341.
Overall, J. E., Hollister, L. E., & Meyer, F. (1964). Imipramine and thioridazine in
depressed and schizophrenic patients. Journal of the American Medical Association,
189, 605-608.
Pelosi, A. & David, A. (1989). Ethical implication of the new genetics for psychiatry.
International Review of Psychiatry, 1, 315-320.
Rachman, S. J. (1981). H. J. Eysenck's contribution to behaviour therapy. In: R. Lynn
(Ed.), Dimensions of personality. Papers in honour of H. J. Eysenck (pp. 315-330).
Oxford: Pergamon.
Raine, A. & Allbutt, J. (1989). Factors of schizoid personality. British Journal of Clinical
Psychology, 28, 31-40.
Raine, A., Lencz, T., & Mednick, S. A. (Eds.) (1995). Schizotypal personality.
Cambridge: Cambridge University Press.
 Eysenck's contribution to understanding psychopathology 387

Rasmussen, K., Levander, S., & Sletvold, H. (1995). Aggressive and non-aggressive
schizophrenics: symptom profile and neuropsychological differences. Psychology,
Crime, and Law, 2, 119-129.
Rice, M. E. & Harris, G. T. (1995). Psychopathy, schizophrenia, alcohol abuse, and
violent recidivism. International Journal of Law and Psychiatry, 18, 333-342.
Rodnight, E. & Gooch, R. N. (1963). A new method for the determination of individual
differences in susceptibility to a depressant drug. In H. J. Eysenck (Ed.), Experiments
with drugs, (pp. 169-193). Oxford: Pergamon.
Roy, A. (Ed.) (1982). Hysteria. Chichester, U.K.: Wiley.
Shagass, C. & Jones, A. L. (1958). A neurophysiological test for psychiatric diagnosis.
American Journal of Psychiatry, 114, 1002-1009.
Sigal, J. J., Star, K. H., & Franks, C. M. (1958). Hysterics and dysthymics as criterion
groups in the study of introversion-extraversion. Journal of Abnormal and Social
Psychology, 57, 143-148.
Slade, P. (1974). Psychometric studies. In: H. R. Beech (Ed.), Obsessional states (pp. 95-
109). London: Methuen.
Stromgren, E. (1994). The unitary psychosis (Einheitpsychose) concept: Past and
present. Neurology, Psychiatry and Brain Research, 2, 201-205.
Swanson, J. (1994). Mental disorder, substance abuse, and community violence: an
epidemiological approach. In J. Monahan & H. Steadman (Eds.), Violence and
mental disorder: Developments in risk assessment (pp. 101-136) Chicago, IL:
University of Chicago Press.
Taylor, M. A. (1992). Are schizophrenia and affective disorder related? A selective
literature review. American Journal of Psychiatry, 149, 22-32.
Williams, J. M. G. (1992). The Psychological Treatment of Depression. (2nd ed.).
London: Routledge.
Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. W. (1988). Cognitive
psychology and emotional disorders. Chichester, U.K.: Wiley.
Zuckerman, M. (1979). Sensation seeking: Beyond the optimal level of arousal. Hillsdale,
NJ: Lawrence Erlbaum.
Zuckerman, M. (1989). Personality in the third dimension: A psychobiological
approach. Personality and Individual Differences, 10, 391-418.
Chapter 18
The psychophysics and psychophysiology of
extraversion and arousal

R. M. Stelmack

1. PERSPECTIVE ON EXTRAVERSION AND AROUSAL

When I was an undergraduate student at the University of Windsor, a chapter

on Factor Theories in the first edition of Hall and Lindzey's (1957) review of
theories of personality provided my first introduction to the work of Hans
Eysenck. The second exposure occurred a few years later when my supervisor
in clinical training, Hugh McLeod, encouraged me to read his doctoral
dissertation on the inheritance of personality that was directed by Professor
Eysenck. At the time, I held more awe than appreciation for factor analysis or
mathematical models so these readings had little impact. My research interests
were taking shape, however, with the objective of applying methods of
experimental psychology to the study of personality. This objective was
expanded to include a conceptual framework when I discovered, after
completing my own doctoral dissertation, that I could not explain the inter-
esting effects that were observed because of the absence of an adequate
personality theory in my work. The recognition of the validity and value of
Eysenck's approach to the study of individual differences was confirmed for me
when my colleague Ken Campbell and I set out to refute the claim that
introverts had lower sensory thresholds than extraverts (Smith, 1968) by using
signal detection procedures that controlled for response bias. The greater
auditory sensitivity of introverts than extraverts was, in fact, confirmed
(Stelmack & Campbell, 1974). From that point, research on the causal bases of
individual differences in extraversion has been one of my main academic
interests.
Publication of The Biological Basis of Personality (Eysenck, 1967) heralded
the transition of the explanatory framework of individual differences in extra-
version from the hypothetical neural constructs of excitation and inhibition to
arousal theory. This transition was abetted by several significant contributions,
notably work by Claridge (1967; see chapter 17) that applied arousal theory to
personality, work by Gray (1964; see chapter 3) that promoted a synthesis of
 The psychophysics and psychophysiology of extroversion and arousal 389

arousal and the Pavlovian typology of the nervous system, and by a burgeoning
literature on the role of activation and arousal mechanisms in attention,
learning, and motivation (e.g., Lindsley, 1958). Together, these developments
provided the impetus for a surge of research that employed a full range of
psychophysical and psychophysiological indices of the arousal construct in the
study of personality.
An extensive body of research on the arousal theory of extraversion amassed
quickly despite important weaknesses in the theory, including: (1) the difficulty
in accounting for some effects in the excitation-inhibition theory, especially
effects involving motor inhibition; (2) the difficulty in demonstrating individual
differences in conditioning, which was a cornerstone of the theory; and (3) by
limitations inherent in the arousal construct itself. Not the least of these
limitations was that the term "arousal" was used by various authors almost
interchangeably with terms such as activation, drive state, and energy
mobilization, often with different measurement indices. Malmo (1959)
advocated a view of arousal that was widely shared. In this view arousal
(activation) was conceived as a dimension that described a continuum of
neuropsychological states, ranging from deep sleep at the low end of the
continuum to excited states at the high end. It was proposed that these states
were a direct function of the amount of cortical bombardment by the ascending
reticular activating system (ARAS) and that the relation between level of
arousal and performance was described by an inverted U-shaped curve.
Although it was recognized in the early 1950s that the ARAS was not a
homogeneous anatomical system, it was argued that the ARAS could function
as a unitary intensity-mediating system. Initially, this function seemed to fit the
requirements of a theory of extraversion quite well especially with respect to
individual differences in perceptual sensitivity, attention, memory, vigilance,
and learning that had been demonstrated. In this view, the ARAS was
understood as a system of sustained activity, like drive, that served to maintain
an optimum level of performance in these functions. There were, however,
some caveats.
As Gray (1981) pointed out, the new arousal theory of extraversion did not
readily accommodate predictions and effects that were borne by the inhibition
hypothesis. The behavioral measures of inhibition employed in the personality
research generally involved a motor movement component such as in finger
tapping and pursuit rotor performance. The extraversion effects observed with
these measures were not readily understood in the context of research on the
ARAS. Second, the arousal construct was under siege in the literature because
the low correlations between various psychophysiological measures did not
support a unitary construct. An antipathy towards the arousal construct was
cultivated that was based largely on the analysis of intercorrelations between
brief psychophysiological responses to punctuate stimuli of the kind employed
in classical conditioning studies. This negative view was widely embraced even
390 Further Eysenckian interests

though the data used to assess the arousal construct were brief reactions to
stimuli, measured in seconds, rather than on sustained indices, measured in
minutes, that were more compatible with the conception of a sustained arousal
state. Malmo (1962) and his co-workers did report a series of studies,
employing electromyographic methods, that were supportive of the effects of
arousal state and with the inverted-U relation of arousal level and
performance. Curiously, these studies were largely overlooked and the
devaluation of the currency of the arousal construct persisted. The antipathy
allocated to the arousal construct also eroded the value of the arousal theory of
extraversion.
There is more than a little irony in the fact that, on the whole, psycho-
physiological analyses of individual differences in extraversion reveal little
compelling evidence of individual differences in levels of arousal state that are
independent of stimulus effects (Stelmack, 1981, 1990). There is, however, a
substantial body of evidence indicating the greater sensory reactivity of
introverts than extraverts to punctuate physiological stimuli of moderate
intensity or "arousal" value. Moreover, this effect is observed across sensory
modalities, with a variety of psychophysical and psychophysiological methods,
and in the expression of individual differences in social behavior. In addition,
there is some psychophysical and psychophysiological research implicating
individual differences in motor processes between introverts and extraverts
that may be relevant to effects observed during early work assessing the
inhibition hypothesis.
In a retrospective volume dedicated to the prime mover of scientific research
on personality, it is appropriate to reconsider the basic objectives of the
research and to consider the extent to which these objectives were realized. In
this brief overview, the specific objective of psychophysical and psychophysio-
logical research on extraversion and arousal is described by focusing on what it
is that research on extraversion attempts to explain. Second, the evidence for
the claim that introverts are more sensitive to punctate physical stimulation
than are extraverts is summarized. Third, some evidence is introduced for the
claim that individual differences in the extraversion trait involve individual
differences in fundamental motor processes.

2. ON UNDERSTANDING EXTRAVERSION
2.1 Introduction
At first glance, the answer to the question "what is it that psychophysical and
psychophysiological research on extraversion is trying to explain?" is
straightforward. Most of the experimental research on extraversion aimed to
assess the arousal hypothesis. That is, the hypothesis that introverts are
characterized by higher levels of cortical arousal than are extraverts. The
arousal construct was indexed by a clutch of psychophysical measures,
 The psychophysics and psychophysiology of extroversion and arousal 391

including visual, auditory, and gustatory thresholds, recognition, and duration

thresholds, and signal detection discriminability indices. In addition, a
complete range of psychophysiological indices were applied, including a full
spectrum of EEG waves, electromyographic measures, almost all of the event-
related potential components, and various electrodermal, cardiac, vascular,
and pupillary autonomic nervous system measures. The plausibility of the
influence of the ARAS on these measures was not questioned; but the extent
of this influence was not at all clear. Moreover, the functional significance of
these indices, especially with respect to individual differences in personality,
was to a great extent impenetrable. As previously noted, the sterile debate
concerning the unitary nature of the arousal construct compounded the
confusion and further clouded the issue. With this slippery grip on the arousal
construct, the primary objective of this work, to explain individual differences
in extraversion, was subordinate to demonstrating differences between groups
in arousal. The conceptual link between extraversion and arousal, however,
was not forged.
At the present time, some reliable and replicable effects can be extracted
from this large body of work that show a good convergence between (1) the
descriptive terms that define the construct of extraversion, (2) the behavioral
expression of the characteristics of extraversion in specific situations, and (3)
some paradigms of experimental psychology that assessed the arousal
hypothesis. These effects do provide a good cornerstone for an explanatory
framework for the extraversion construct and perhaps provide some direction
in establishing experimental procedures for further exploration of the
biological basis of extraversion.

2.2 A brief note on the description and validity of the extraversion trait
Extraversion is a descriptive construct that is defined by the common factor
loadings of test items. Salient items in the Eysenck Personality Questionnaire
(H. J. Eysenck & S. B. G. Eysenck, 1975) by which extraverts describe them-
selves include sociable (like to mix with others), socially uninhibited (can let
themselves go at a lively party), lively and active (others think of you as lively),
talkative, and socially dominant (can get a dull party going; take initiative in
meeting new friends). Whereas introverts describe themselves as reserved
(mostly quiet when others are around) and socially distant (stay in the back-
ground when other people are around). Thus, the extraversion trait is present-
ed as a disposition to engage in social stimulation (especially communicative,
talkative) and to indulge in social activity (especially liveliness, parties, sports).
In developing the meaning of this trait here, it may be useful to think of social
stimulation as involving stimulation of all of the senses (sights, sounds, smells)
with extraverts being more tolerant and introverts more susceptible to the
stimulation and also to think of social activity in terms where extraverts are
392 Further Eysenckian interests

more physically active (lively, talking, moving, etc.), and where introverts adopt
more contemplative poses. It is variation in this behavioral disposition that is
described which requires explanation.
Evidence that these self-descriptions of the extraversion trait are valid is
provided in several ingenious studies. Although, it is somewhat surprising that
the total number of these validation studies is rather small. These data do
provide behavioral descriptions of the extraversion trait that are extensions, or
validations, of the test items and that are objective descriptions of the trait as it
is expressed in specific situations. There are several studies that report the
performance of introverts and extraverts in experimental situations which
confirm that introverts prefer quieter, less stimulating environments with fewer
socializing opportunities than do extraverts. Campbell and Hawley (1982)
demonstrated that, in contrast to extraverts, introverts chose to study in
quieter, less crowded locations in the library that had lower levels of visual and
auditory stimulation. Several subsequent studies on this subject also strongly
supported the finding that introverts are more sensitive and have less tolerance
to stimulation than extraverts (Campbell, 1983, 1992; Domic & Ekehammar,
1990). In another impressive study, it was demonstrated that extraverts set the
intensity of sound almost 20 dB higher than introverts, 70 versus 54 dB, when
given free choice of intensity level (Geen, 1984).
That extraverts are more talkative than introverts was demonstrated in a
study that examined communication during an interview situation (Campbell
& Rushton, 1978). It was observed that extraverts initiated the conversation
more frequently and they were more talkative during the interview than
introverted participants. Studies that assessed the personalities of athletes have
observed that athletes tend to be more extraverted than population norms
(Eysenck, Nias, & Cox, 1982; Kirkcaldy, 1982). This observation endorses the
view that extraverts are more disposed to physical activity than introverts.
Some insight into the causal bases of these behaviors is provided by
psychophysical and psychophysiological research that examined individual
differences in sensory and motor processes between introverts and extraverts.

3. EXTRAVERSION AND SENSITIVITY TO STIMULATION

3.1 Sensory psychophysical evidence
A number of research reports were published during the 1960s and 1970s that
assessed the sensory sensitivity of introverts and extraverts using psychophy-
sical methods. The rationale for some of these studies was derived from the
excitation-inhibition hypothesis, but the effects predicted were quite consistent
with the arousal theory. The greater sensitivity of introverts than extraverts to
low-intensity stimulation has been observed in both auditory (Smith, 1968;
Stehnack & Campbell, 1974) and visual modalities (Siddle, Morrish, White,
Mangan, 1969). Individual differences in the effects of cross-modality
 The psychophysics and psychophysiology of extroversion and arousal 393

stimulation were also investigated. Shigehisa and Symons (1973) reported that
the auditory sensitivity of introverts increased when stimulated with weak-
intensity light pulses, but decreased under strong light. Whereas, auditory
sensitivity of extraverts increased when stimulated with a full range of weak,
moderate, and strong light pulses. Similar differences between groups were
observed when the effects of auditory stimulation on visual thresholds were
examined (P. T. Shigehisa, T. Shigehisa, & Symons, 1973). In both studies,
introverts were more susceptible to the disruptive effects of moderate and
high-intensity stimulation. There is also a small body of literature that
determined pain thresholds for introverts and extraverts which show similar
results, with introverts providing lower pain thresholds than extraverts in a
clear majority of studies (Barnes, 1975; Dubrieul & Kohn, 1986; H. J. Eysenck
& M. W. Eysenck, 1985). Although, the psychophysical literature is rather
limited, the overall direction of effects is quite clear, in that introverts appear
more sensitive to punctate physical stimulation than extraverts across all
modalities, for both high and low intensity, and in their preference for lower
intensity stimulation than extraverts. It is also interesting to note that
psychophysical discrimination thresholds, which are inversely related to
intelligence, do not correlate with extraversion.

3.2 Sensory psychophysiological evidence

There is a substantial body of work examining individual differences in
extraversion using psychophysiological methods that has been recently
reviewed (Stelmack, 1990; Stelmack & Geen, 1992; Stelmack & Houlihan,
1995). The conclusions from those reviews will be briefly noted here. First,
from studies using electrodermal measures, there is a good deal of evidence
demonstrating that introverts exhibit greater skin conductance response (SCR)
amplitudes to moderate intensity tones, that is, 75-90 dB, than do extraverts.
Higher or lower intensity stimulation tends to result in reverse or null effects.
Similar effects are observed with event-related potential (ERP) measures, with
higher amplitude response for introverts than extraverts generally observed to
moderate intensity stimulation. In both the electrodermal and event-related
potential work, these effects tend to manifest themselves more clearly in the
first trials of a series (e.g., Stelmack & Michaud-Achorn, 1985) and under
conditions of moderate-intensity stimulation where individual variation in the
dependent variable is greatest. That the effects emerge early in stimulation
sequences is also congruent with observations made by Koelega (1992) in his
meta-analysis of extraversion and variation in vigilance performance.
With respect to the SCR and ERP studies, it is difficult to attribute the
amplitude effects to anything other than an enhanced response to stimulation
for introverts. Response amplitudes increase monotonically with increases in
stimulus intensity for both measures. Typically, subjects sit passively receiving
394 Further Eysenckian interests

the stimulation and there is virtually no evidence of differences in attentional

set in these studies that could account for such effects. Also, as the Koelega
(1992) review demonstrates, the evidence of behavioral differences in
sustained attention between introverts and extraverts is also very limited.
Moreover, in several studies faster response latency of brainstem auditory
evoked potential (BAEP) components for introverts than extraverts were
observed (Bullock & Gilliland, 1993; Stelmack & Wilson, 1982; Szelenberger,
1983) and there is some evidence that these effects may be observed even when
subjects are asleep (Stelmack, Campbell, & Bell, 1993). BAEP waves develop
in response to auditory click stimuli within the first 10 ms following
stimulation. The latency of BAEP waves vary inversely with increases in
stimulus intensity and the latency is not influenced by variation in attention,
sleep, arousal, or metabolic coma so long as the brainstem structure is intact.
The faster latency of BAEP components for introverts means that greater
auditory sensitivity, or reactivity to stimulation, is evident at the level of the
auditory nerve. These effects implicate differences in peripheral nervous
system processes that are not determined by mechanisms in the ARAS as
proposed in the arousal hypothesis (Eysenck, 1967). The BAEP effects are
significant because they would require an elaboration of the neurological bases
of extraversion to accommodate differences in neuronal transmission that are
present in peripheral nervous system processes.
Some psychophysiological work using electromyographic measures also
merits some mention. Blumenthal and his colleagues have conducted a series
of studies recording the eye blink startle reflex that have provided findings that
are compatible with the intensity effects described above. Startle reflex
response amplitude increases, and response latency decreases with increase in
intensity of the eliciting stimulus. Introverts exhibited faster reflex response
latencies to 85 dB than 60 dB noise bursts, whereas extraverts did not (Britt &
Blumenthal, 1992). Thus introverts, but not extraverts were differentially
sensitive to the increase in stimulus intensity. In subsequent work, social
encounter stimulation moderated reflex effects for introverts but not extraverts
(Muse & Blumenthal, 1995). Using higher intensity noise bursts (100 dB),
Kumari et al. (1996) observed higher baseline EMG levels for introverts than
extraverts in the startle paradigm. Overall, the startle reflex appears to be a
simple procedure that is sensitive to individual differences in extraversion. It
promises to serve as a useful model for probing the neural basis of extraversion
as well as for exploring and verifying characteristics that are thought to
distinguish introverts and extraverts.
Overall, the psychophysical and psychophysiological work that has
accumulated over many years, provides compelling evidence for the claim
that introverts are characterized by greater reactivity to punctate physical
stimuli than are extraverts, and that this effect must be accommodated as a
fundamental fact in the causal basis of extraversion.
 The psychophysics and psychophysiology of extroversion and arousal 395

4. EXTRAVERSION AND MOTOR EXPRESSION

4.1 Introduction
The greater sensitivity and reactivity to physical stimulation that distinguishes
introverts from extraverts can account for some of the social behavior exhibited
by introverts, notably preferences for quiet environments and solitude.
However, these effects do not capture differences in activity, talkativeness,
and spontaneity that are also known to distinguish introverts and extraverts
and that could be understood more readily in terms of differences in
mechanisms that mediate the expression of motor behavior. As previously
noted, some studies assessing Eysenck's inhibition hypothesis compared
introverts and extraverts on motor tasks. Another series of studies employing
motor tasks assessed the hypothesis that introverts and extraverts differ in
excitation and inhibition processes derived from response organization
(Brebner & Cooper, 1974). There is also a limited amount of research that
has employed psychophysiological procedures.

4.2 Extroversion and motor performance

Individual differences in motor performance between introverts and extraverts
have been examined in a number of different ways. In reaction time tasks, the
outcomes are mixed, with several studies reporting faster reaction time for
extraverts than introverts (Barratt, 1959, 1967; Buckalew, 1973; Keuss &
Orlebeke, 1977; Robinson & Zahn, 1988) and but others reporting null effects
(Casal, Caballo, Cueto, & Cubos, 1990; Gupta & Nicholson 1985, Hummel &
Lester, 1977; Kirkcaldy, 1987). It does appear, though, that when significant
effects are observed, extraverts tend to have faster response times than
introverts.
Individual differences between introverts and extraverts were also observed
during the pursuit-rotor tracking task. The results of several studies were
reviewed by Frith (1971). The results are complex but the salient effect is that
extraverts perform better following a brief rest pause than do introverts (Frith;
Horn, 1975). Different movement strategies appear to play a role on this task.
Introverts adopted a strategy that appeared to involve the frequent analysis of
the relative positions of the target and marker stimuli. Extraverts, on the other
hand, tended to organize a continuous movement and when velocity matching
was denied, they were less accurate than introverts (Eysenck & Frith, 1977;
Shamberg, Baker, & Burns, 1969).
Differences in motor performance have surfaced in several other paradigms
too. Extraverts tend to make more false positive errors in reaction time tasks
(Brebner & Flavell, 1978), to respond faster in a copying task (Farley, 1966),
and to be more expansive than introverts in free writing and drawing tasks
(Taft, 1967; Wallach & Gahm, 1960). In our own work, we have observed that
396 Further Eysenckian interests

extroverts exhibit faster movement time in cognitive decision tasks than

introverts (Stelmack, Houlihan, & McGarry-Roberts, 1993). Specifically, there
were no differences between groups in reaction time, defined as the time from
target stimulus onset to the release of a home button, but rather in the time
required to press the response button following release of the home button. In
recent work, a simple reaction-time experiment was conducted in which
movement time was varied by placing response buttons at 7, 15, and 23 cm
distant from a home button (Doucet & Stelmack, 1997). There were no
differences in reaction time, but again extroverts exhibited faster movement
time than introverts across all three distances. It was also observed that the
magnitude of the difference between introverts and extroverts across the three
distances was the same, indicating that the movement time differences between
groups was due to the initial movement phase rather than developing during
increasing ballistic movement phase. In this same study, faster movement time
for introverts than extroverts was observed in four conditions that were
composed of congruent and incongruent stimulus patterns and response
demands which were compatible or incompatible with information given in the
stimulus pattern.
Overall, this work does seem to point to basic differences between introverts
in the performance of motor tasks. The review of this work leads me to believe
that the effects may be due to the faster initiation rather than the completion of
movement. Intuitively, this view seems compatible with the spontaneity, both
active and social, that distinguishes extroverts from introverts.

4.3 Extroversion and psychophysiological motor response

There are several psychophysiological procedures that can be used to assess
motor processes. Three that have been employed to study individual
differences in extroversion are: (1) the startle (blink) reflex; (2) the contingent
negative variation (CNV) which is an event-related potential component that
provides a useful index of motor preparation; and (3) the Hoffman reflex (H-
reflex) which is an electromyographic effect that has been used to index spinal
motoneuronal recovery. As previously noted, the startle reflex effects that
distinguish introverts and extroverts appear to be determined by the greater
sensitivity to stimulation of introverts rather than by differences hi the startle
reflex itself.
There is no significant CNV work comparing introverts and extroverts that
can be added to studies reviewed 12 years ago by Stelmack (1985). Smaller
CNV amplitude for introverts, indicating less effective response preparation,
was observed in several studies. In the only study which used an interval
between the warning stimulus and the imperative stimulus to respond that
permitted a clear distinction to be made between orienting to the warning
stimulus and expectancy of the imperative stimulus, smaller amplitude for
 The psychophysics and psychophysiology of extroversion and arousal 397

introverts than extraverts was observed (Plooij-Van Gorsel & Janssen, 1978).
In another study by these authors, however, smaller CNV amplitude for
introverts was only observed in a noise condition (Janssen, Mattie, Plooij-van
Gorsel, Werre, 1978). There are several reports indicating that nicotine has the
effect of increasing CNV amplitude for extraverts and decreasing CNV
amplitude for introverts (e.g., O'Connor, 1982). This is an intriguing effect, in
particular because of speculation that at smoking doses, nicotine appears to
mimic acetylcholine and may induce neural changes that are similar to those
produced by the natural activation of cholinergic synapses (Edwards &
Warburton, 1983). The view that there may be differences in acetylcholine
release does fit with the sensory and motor effects that distinguish introverts
and extraverts. In general, however, it is not clear from this work whether the
CNV effects observed reflect differences in motor preparation, sensory sensi-
tivity or both.
Individual differences in spinal motoneuronal excitability between introverts
and extraverts were examined in two studies that recorded the spinal
monosynaptic H-reflex using electromyographic methods. The H-reflex was
elicited by brief electrical pulses that stimulate sensory fibers from muscle
spindle receptors, and it was recorded as the amplitude of the evoked reflex
muscle action potential. Pairs of electrical pulses were presented at varying
interstimulus intervals (50 ms to 2 s). Reflex recovery was expressed as a ratio
of the reflex amplitude of the second stimulus in the pair to the first stimulus in
the pair. In both studies, extraverts exhibited less reflex recovery than
introverts (Pivik, Stelmack, & Bylsma, 1988; Stelmack & Pivik, 1996). The
personality groups did not differ in the intensity of the stimulation required to
elicit muscle action potentials or the nerve conduction velocity of those
potentials. Thus, these effects were not attributable to differences in initial
levels of excitability, but rather the excitability changes that became evident
once motoneuronal activity was initiated.
It can be argued that these effects may be indicative of greater dopaminergic
activity for extraverts because lower amplitude recovery functions are related
to increased dopaminergic activity (Goode & Manning, 1984). The H-reflex
recovery function is reduced in amplitude by dopamine agonists (Goode,
Meltzer, & Mazura, 1979) and augmented by dopamine receptor blockers
(Meltzer & Stahl, 1976). Moreover, there is some evidence that differences in
motoneuronal excitability, as indexed by H-reflex amplitude, differentiate
healthy subjects from clinical groups, that is, psychotic patients (Crayton,
Meltzer, & Goode, 1977) and hyperkinetic children (Pivik, Bylsma, &
Margittai, 1986; Pivik & Mercier, 1981) that are thought to be characterized
by abnormalities in catecholamine systems.
Direct evidence associating higher extraversion scores with increased
dopaminergic activity is equivocal. An inverse relation between extraversion
and levels of monoamine oxidase and dopamine-beta-hydroxylase was
398 Further Eysenckian interests

observed (Murphy et al., 1977). These neurochemical variations could be

associated with increased dopaminergjc activity. In two large-scale studies,
however, null results were reported (von Knorring, Oreland, & Winblad, 1984;
Zuckerman, Ballenger, Jimerson, Murphy, & Post 1983).
In a recent study, dopaminergic activity was directly manipulated by
administering a dopamine inhibitor, a-methyl-p-tyrosine (AMPT) and
observing the effects on introverts and extraverts during the performance of
a choice reaction-time task (Rammsayer, Netter, & Vogel, 1993). AMPT
reduces the synthesis of dopamine by inhibiting tyrosine hydroxylase. It had the
effect of increasing reaction time and movement time for introverts but had no
effect on the extravert group. The authors argued that the effects described
were not determined by differences in dopaminergic states or levels, but
possibly to differences in postsynaptic receptor sensitivity. The authors put
forth a cogent rationale for the study that was based on a model for action of
the mesostriatal dopamine system (Heffner, Zigmond, & Strieker, 1977). This
model assumes a regulatory role for dopaminergic activity (DA) in which DA
modulates the probability and strength of behavioral responses to sensory
input (Le Moal & Simon, 1991). In view of the clear convergence of results
between descriptive, behavioral, and experimental methods, indicating that
differences in the extraversion trait involve individual differences in
fundamental sensory and motor mechanisms, such a sensory-motor
hypothesis deserves a good deal of consideration as a plausible basis for
individual differences in extraversion.

5. RETROSPECTIVE ON EXTRAVERSION AND AROUSAL

The field of personality is only one of several areas to which Hans Eysenck has
made substantial, foundational contributions. His theoretical work has had
enormous heuristic influence on thousands of students and scholars who have
labored with him to compose a valid empirical basis for individual differences
in social and personal behavior and in their expression in psychiatric disorders.
My own work on personality, which has benefited greatly from his support and
leadership, has focused to a large extent on the extraversion trait. This trait is
now firmly established as a fundamental construct in personality typology.
There is also compelling evidence demonstrating that, to a large extent, this
trait is heritable. Because of this finding, there is good reason to believe that
the biological basis of extraversion can be found.
The excitation-inhibition and arousal theories of extraversion that Eysenck
put forth were drawn from contemporary approaches to learning and
motivation. Individual differences in sensory sensitivity and in the expression
of motor behavior were implicated in both of these theories. In my opinion,
much of the work that was spawned by the theories does endorse the view that
 The psychophysics and psychophysiology of extraversion and arousal 399

introverts are more sensitive to physical stimulation than are extraverts and
that differences in the extraversion trait involve individual differences in
fundamental motor mechanisms. There does not, however, appear to be a
strong case, at any level of analysis, indicating that these effects are determined
by differences in cortical arousal state as originally conceived. The sensory
sensitivity and motor response effects that distinguish introverts and extraverts
appear to emerge early and briefly in the processing of both sensory and motor
information. Also, evidence suggesting that differences in extraversion may
involve peripheral brainstem and spinal motoneuronal processes challenges
the view that differences in extraversion are determined primarily by central
cortical arousal mechanisms. In view of the regulatory role that dopaminergic
activity plays in modulating sensory input and response output, variation in
dopaminergic activity may prove to be an important determinant of individual
differences in extraversion. Even if such a view does prevail, however,
Eysenck's original conception is really not that far from the mark.

REFERENCES

Barratt, E. (1959). Anxiety and impulsiveness related to psychomotor efficiency.

Perceptual and Motor Skills, 9, 191-198.
Barnes, G. (1975). Extraversion and pain. British Journal Of Social And Clinical
Psychology, 14, 303-308.
Barratt, E. (1967). Perceptual-motor performance related to impulsiveness and anxiety.
Perceptual and Motor Skills, 25, 485-492.
Brebner, J. & Cooper, C. (1974). The effects of a low rate of regular signals upon the
reaction time of introverts and extraverts. Journal of Research in Personality, 8, 263-
276.
Brebner, J. & Flavell, R. (1978). The effect of catch-trials on speed and accuracy among
introverts and extraverts in a simple RT task. British Journal of Psychology, 69, 9-15.
Britt, T. W. & Blumenthal, T. D. (1992). The effects of anxiety on motoric expression of
the startle response. Personality and Individual Differences, 13, 91-98.
Buckalew, L. W. (1973). Relationship between a physiological and personality index of
excitability. Physiological Psychology, 1, 158-160.
Bullock, W. A. & Gilliland, K. (1993). Eysenck's arousal theory of introversion-
extraversion: A converging measures investigation. Journal of Personality and Social
Psychology, 64, 113-123.
Campbell, J. B. (1983). Differential relationships of extraversion, impulsivity, and
sociability to study habits. Journal of Research in Personality, 17, 308-314.
Campbell, J. B. (1992). Extraversion and noise sensitivity: A replication of Dornic and
Ekehammar's study. Personality and Individual Differences, 13, 953-956.
Campbell, J. B. & Hawley, C. W. (1982). Study habits and Eysenck's theory of
extraversion-introversion. Journal of Research in Personality, 16, 139-146.
Campbell, A. & Rushton, J. P. (1978). Bodily communication and personality. British
Journal of Social and Clinical Psychology, 17, 31-36.
400 Further Eysenckian interests

Casal, G. B., Caballo, V. E., Cueto, E. G. & Cubes, P. F. (1990). Attention and reaction
time differences in introversion-extraversion. Personality and Individual Differences,
11, 195-197.
Claridge, G. S. (1967). Personality and arousal. Oxford: Pergamon.
Crayton, J. W., Meltzer, H. Y., & Goode, D. J. (1977). Motoneuronal excitability in
psychiatric patients. Biological Psychiatry, 12, 545-561.
Doucet, C. & Stelmack, R. M. (in press). Movement time differentiates extraverts from
introverts. Personality and Individual Differences, 21.
Domic, S. & Ekehammar, B. (1990). Extraversion, neuroticism, and noise sensitivity.
Personality and Individual Differences, 11, 989-992.
Dubreuil, D. L. & Kohn, P. M. (1986). Reactivity and response to pain. Personality and
Individual Differences, 7, 907-910.
Edwards, J. A. & Warburton, D. M. (1983). Smoking, nicotine and electrocortical
activity. Psychopharmacology Therapy, 19, 147-164.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: Thomas.
Eysenck, H. J. & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality
Questionnaire. London: Hodder and Stoughton Educational.
Eysenck, H. J. & Eysenck, M. W. (1985). Personality and Individual Differences. New
York: Plenum Press.
Eysenck, H. J. & Frith, C. D. (1977). Reminiscence, motivation and personality, New
York: Plenum Press.
Eysenck, H. J., Nias, D. K. B., & Cox, D. N. (1982). Sport and personality. Advances in
Behavior Research and Therapy, 4, 1-56.
Farley, F. (1966). Individual differences in solution time in error-free problem solving.
British Journal of Social and Clinical Psychology, 5, 306-309.
Frith, C. D. (1971). Strategies in rotary pursuit tracking. British Journal of Psychology,
62, 187-197.
Geen, R. G. (1984). Preferred stimulation levels in introverts and extraverts: Effects
on arousal and performance. Journal of Personality and Social Psychology, 46,
1303-1312.
Goode, D. J., Meltzer, H. Y., & Mazura, T. A. (1979). Hoffman reflex abnormalities in
psychiatric patients. Biological Psychiatry, 14, 95-110.
Goode, D. J. & Manning, A. A. (1984). Variations in the Hoffman recovery curve
related to clinical manifestations of schizo-affective disorder. Psychiatry Research, 15,
63-70.
Gray, J. A. (1964). Strength of the nervous system and levels of arousal: a reinter-
pretation. In J. A. Gray (Ed.), Pavlov's typology (pp. 289-366). Oxford: Pergamon.
Gray, J. A. (1981). A critique of Eysenck's theory of personality. In H. J. Eysenck (Ed.),
A model for personality (pp. 246-276). Berlin: Springer.
Gupta, A. & Nicholson, J. (1985). Simple visual reaction time, personality and strength
of the nervous system: A signal detection theory approach. Personality and Individual
Differences, 6, 461^469.
Hall, C. S. & Lindzey, G. (1957). Theories of personality. New York: Wiley.
Heffher, T. G., Zigman, M. J. & Strieker, E. M. (1977). Effects of dopaminergic agonists
and antagonists on feeding hi intact and 6-hydroxydopamine treated rats. Journal of
Pharmacology and Experimental Therapeutics, 14, 380-399.
 The psychophysics and psychophysiology of extroversion and arousal 401

Horn, P. D. (1975). Evidence for the generality of reminiscence function of extraversion

and neuroticism. Journal of Psychology, 90, 41-44.
Hummel, H. & Lester, D. (1977). Extraversion and simple reaction time. Perceptual and
Motor Skills, 45, 1236.
Janssen, R. H. C, Mattie, H., Plooij-van Gorsel, P. C, & Werre, P. F. (1978). Biological
Psychology, 6, 209-218.
Keuss, P. J. G. & Orlebeke, J. F. (1977). Transmarginal inhibition in a reaction time
task as a function of extraversion and neuroticism. Acta Psychologica, 41, 139-150.
Kirkcaldy, B. D. (1982). Personality profiles at various levels of athletic participation.
Personality and Individual Differences, 3, 321-326.
Kirkcaldy, B., (1987). The relationship between personality variables and performance
in a clinical sample. Studia Psychologica, 29, 227-232.
von Knorring, L., Oreland, L. & Winblad, B. (1984). Personality traits related to
monoamine oxidase (MAO) activity in platelets. Psychiatry Research, 13, 11-26.
Koelega, H. S. (1992). Extraversion and vigilance performance: 30 years of
inconsistencies. Psychological Bulletin, 112, 239-258.
Kumari, V., Corr, P. J., Wilson, G. D., Kaviani, H., Thornton, J. C., Checkley, S. A., &
Gray, J. A. (1996). Personality and modulation of the startle reflex by emotionally-
toned filmclips. Personality and Individual Differences, 21, 1029-1041.
Le Moal, M. & Simon, H. (1991). Mesocorticolimbic dopaminergic network: Functional
and regulatory roles. Physiological Reviews, 71, 155-234.
Lindsley, D. B. (1958). Psychophysiology and motivation. In M. R. Jones (Ed.),
Nebraska symposium on motivation (pp. 44-105). Lincoln: University of Nebraska
Press.
Malmo, R. B. (1959). Activation: A neurophysiological dimension. Psychological Review,
8, 367-386.
Malmo, R. B. (1962). Activation. In A. J. Bachrach (Ed.), Experimental foundations of
clinical psychology (pp. 386-422). New York: Basic Books.
Meltzer, H. Y. & Stahl, S. M. (1976). The dopamine hypothesis of schizophrenia: A
review. Schizophrenia Bulletin, 2, 19-76.
Murphy, D. L., Belmaker, R. H., Buchsbaum, M. S., Martin, N. F., Ciaranello, R. &
Wyatt, R. J. (1977). Biogenic amine related enzymes and personality variations in
normals. Psychological Medicine, 7, 149-157.
Muse, K. B. & Blumenthal, T. D. (1995). Effects of social encounter on startle
responding in introverts and extraverts. Psychophysiology (Suppl. 1), 32, 54.
O'Connor, K. (1982). Individual differences in the effect of smoking on frontal-central
distribution of the CNV: Some observations on smokers' control of attentional
behaviour. Personality and Individual Differences, 3, 271-286.
Pivik, R. T., Bylsma, F. W. & Margittai, K. (1986). Abnormal motoneuronal excitability
in hyperkinesis: H-reflex recovery function and homosynaptic depression during
wakefulness. Psychophysiology, 23, 146-155.
Pivik, R. T. & Mercier, L. (1981). Spinal motoneuronal excitability in hyperkinesis: H-
reflex recovery function and homosynaptic depression during wakefulness. Journal of
Clinical Neuropsychology, 3, 215-236.
Pivik, R. T., Stelmack, R. M., & Bylsma, F. W. (1988). Personality and individual
differences in spinal motoneuronal excitability. Psychophysiology, 25, 16-24.
402 Further Eysenckian interests

Plooij-van Gorsel, P. C. & Janssen, R. H. C. (1978). Contingent negative variation

(CNV) and extraversion in a psychiatric population. In C. Barber (Ed.), Evoked
potentials: Proceedings of an international evoked potential symposium, Nottingham
University, (pp. 505-514). Lancaster, U.K.: MTP Press.
Rammsayer, T., Netter, P., Vogel, W. H. (1993). A neurochemical model underlying
differences in reaction times between introverts and extraverts. Personality and
Individual Differences, 14, 701-712.
Robinson, T. N. & Zahn, T. P. (1988). Preparatory interval effects on the reaction time
performance of introverts and extraverts. Personality and Individual Differences, 9,
749-761.
Shamberg, N., Baker, S. & Burns, J. (1969). Reminiscence and pursuit-rotor
performance in introverts and extraverts. British Journal of Social and Clinical
Psychology, 8, 375-382.
Shigehisa, T. & Symons, J. (1973). Effects of intensity of visual stimulation on auditory
sensitivity in relation to personality. British Journal of Psychology, 64, 205-213.
Shigehisa, P. T., Shigehisa, T., & Symons, J. R. (1973). Effects of intensity of auditory
stimulation on photopic visual sensitivity in relation to personality. Japanese
Psychological Research, 15, 164-172.
Siddle, D. A. T., Morrish, R. B., White, K. D., & Mangan, G. L. (1969). Relation of
visual sensitivity to extraversion. Journal of Experimental Research in Personality, 3,
264-267.
Smith, S. L. (1968). Extraversion and sensory thresholds. Psychophysiology, 5, 296-297.
Stelmack, R. M. (1981). The psychophysiology of extraversion and neuroticism. In H. J.
Eysenck (Ed.), A model for personality (pp. 38-64). Heidelberg, Germany: Springer.
Stelmack, R. M. (1985). Personality and motor activity: A psychophysiological
perspective. In B. Kirkcaldy (Ed.), Individual differences in movement (pp. 193-
213). Lancaster, U.K.: MTP.
Stelmack, R. M. (1990). The biological basis of extraversion: Psychophysiological
evidence. Journal of Personality, 58, 293-311.
Stelmack, R. M. & Campbell, K. B. (1974). Extraversion and auditory sensitivity to high
and low frequency. Perceptual and Motor Skills, 38, 875-879.
Stelmack, R. M., Campbell, K. B., & Bell, I. (1993). Extraversion and brainstem
auditory evoked potentials during sleep and wakefulness. Personality and Individual
Differences, 14, 447-453.
Stelmack, R. M. & Geen, R. G. (1992). The psychophysiology of extraversion. In A.
Gale & M. W. Eysenck (Eds.), Handbook of individual differences: Biological
perspectives (pp. 227-254). New York: Wiley.
Stelmack, R. M. & Houlihan, M. (1995). Event-related potentials, personality and
intelligence: Concepts, issues and evidence. In D. H. Saklofske & M. Zaidner (Eds.),
International handbook of personality and intelligence (pp. 349-365). New York:
Plenum.
Stelmack, R. M., Houlihan, M., & McGarry-Roberts, P. A. (1993). Personality, reaction
time, and event-related potentials. Journal of Personality and Social Psychology, 65,
399-409.
Stelmack, R. M. & Michaud-Achorn, A. (1985). Extraversion, attention, and
habituation of the auditory evoked response. Journal of Research in Personality, 19,
416^28.
 The psychophysics and psychophysiology of extroversion and arousal 403

Stelmack, R. M. & Pivik, R. T. (1996). Extraversion and the effects of exercise on spinal
motoneuronal excitability. Personality and Individual Differences, 21, 69-76.
Stelmack, R. M. & Wilson, K. G. (1982). Extraversion and the effects of frequency and
intensity on the auditory brainstem evoked response. Personality and Individual
Differences, 3, 373-380.
Szelenberger, W. (1983). Brain-stem auditory evoked potentials and personality.
Biological Psychiatry, 18, 157-174.
Taft, R. (1967). Extraversion, neuroticism and expressive behavior: An application of
Wallach's moderator effect to handwriting analysis. Journal of Personality, 35, 570-
584.
Wallach, M. A. & Gahm, R. C. (1960). Personality functions of graphic construction
and expansiveness. Journal of Personality, 28, 73-80.
Zuckerman, M., Ballenger, J. C., Jimerson, D. C., Murphy, D. L., & Post, R. M. (1983).
A correlational test in humans of the biological models of sensation seeking,
impulsivity and anxiety. In M. Zuckerman (Ed.), Biological basis of sensation seeking,
impulsivity, and anxiety (pp. 229-248). Hillsdale, NJ: Erlbaum.
Chapter 19
(Im)pure genius—psychoticism, intelligence,
and creativity

/. P. Rushton
1. INTRODUCTION

A defining feature of great creativity is its statistical rarity, which poses a

problem for purely sociocultural explanations. While sociocultural theorists
might claim that the appearance of the theory of evolution by natural selection
became inevitable in the middle of the nineteenth century, no-one claims that
the Fifth Symphony would have emerged in the early 1800s whether or not
Ludwig van Beethoven existed. Moreover, most lists of "multiple discoveries"
(required by Zeitgeist theories) turn out, on examination, to be quite short and
do not take notice of important individual contributions (Simonton, 1988).
Because Darwin's theory was not identical to Wallace's, the course of
biological thought would likely have been very different had Darwin drowned
while on the Beagle voyage.
In his masterwork Genius: The Natural History of Creativity, Hans Eysenck
(1995) proposes that some individuals are more creative than others because
they are higher in psychoticism, having a relative excess of dopamine and a
relative deficit of serotonin. A moderate degree of psychoticism involves wide
associative horizons and overinclusive thinking which facilitate the discovery of
remote associations, which is the basis for creative inspiration. Add product-
ivity to creativity and you get achievement, with the term "genius" reserved for
work of outstanding achievement. In this chapter I will focus mainly on
achievement in science, especially in psychology, where a number of publi-
cations and citations (scholarly impact) provide objective indices. Publications
require at least a minimum of creativity and large citation counts suggest
methodological and theoretical advances.

2. SCIENTIFIC ACHIEVEMENT

It is generally agreed that whereas personality and intelligence are normally

distributed, scientific achievement is very abnormally distributed. Only a
 (Im)pure genius—psychoticism, intelligence, and creativity 405

relatively few active scientists are responsible for the great majority of creative
works. Across disciplines, Dennis (1955) and Shockley (1957) found that the
most productive 10% of scientists accounted for 50% of the publications,
whereas the least productive 50% accounted for only 15% of the publications.
These figures actually underestimate the differences because they only include
those who published at least one paper, leaving out of consideration those
never making any contribution at all!
Studies of academic psychologists have taken zero producers into account
(Endler, Rushton, & Roediger, 1978; Rushton, 1989). Consider, for example,
the citation and publication counts reported in Table 19.1. These cumulative
percentage frequencies are based on 4070 faculty members studied by Endler
et al. in an analysis of the top 100 departments of psychology in the U.S.A.,
Canada, and the U.K. From Table 19.1 it can be seen that 52% of the sample

Table 19.1. Frequencies and cumulative percentage frequencies for the distribution of citations of
and publications by faculty members at the top 100 British, Canadian, and American graduate
departments of psychology

Citations Publications

Number of Cumulative Cumulative

citations or percentage percentage
publications Frequency frequency Frequency frequency
>100 134 100 —
26-99 556 97 —
21-25 164 83 1 100
16-20 223 79 1 99
11-15 338 74 1 99
10 97 65 3 99
9 82 63 4 99
8 102 61 12 99
7 105 58 18 99
6 125 56 37 99
5 187 53 54 98
4 187 48 147 97
3 207 44 259 93
2 302 38 468 87
1 365 31 971 75
0 896 22 2094 52
Total 4070 4070

Note. From the 1975 Social Sciences Citation Index. (From Endler, Rushton, and Roediger, 1978,
p. 1079, Table 5.) Copyright 1978 by the American Psychological Association. Reprinted by
permission.
406 Further Eysenctdan interests

did not publish an article in 1975 in any of the journals reviewed by the Social
Sciences Citation Index. The picture is similar for citations, the great majority of
academic psychologists having relatively few. For example, only about 25% of
psychologists had more than 15 citations in 1975 and only 1% had more than
100 citations.
"Agist," "sexist," and "elitist" factors contribute to the positive skew in the
distributions. Studies find that productivity increases with age up to around 40-
45 years and then gradually diminishes; that women are not only under-
represented in science but, on a per capita basis, produce less than their male
counterparts; and that individuals who receive doctorates from more
prestigious institutions and/or who get then" first academic positions at high-
prestige universities are more productive than those who graduate from or are
appointed to less esteemed institutions.
As Walberg, Strykowski, Rovai, & Hung (1984) explain, the normal distri-
bution does not apply to exceptional performance. Instead, J-shaped dist-
ributions such as those shown hi Table 19.1, are characteristic. J-shaped
distributions—monotonically decreasing at a decelerating rate—typically arise
when the underlying causes combine multiplicatively rather than additively.
(Additive causes typically produce normal distributions.) Walberg et al. show
that for education, learning is a multiplicative, diminishing-returns function of
student ability, time, motivation, and amount and quality of instruction (those
instances in which no learning at all takes place occur because any zero score in
the equation yields a product of zero).

3. EYSENCK'S THEORY OF CREATIVITY

3.1 Introduction
Eysenck (1995) elaborated on Walberg et al. (1984) and suggested that creative
achievement is a multiplicative function of cognitive, personality, and
environmental variables as shown in Figure 19.1. Cognitive abilities (such as
intelligence, acquired knowledge, technical skills, and special talents) combine
with personality traits (such as internal motivation, confidence, nonconformity,
persistence, and originality) and environmental variables (such as political-
religious, socioeconomic, and educational factors) to produce truly creative
achievements. Many of these variables are likely to act in a multiplicative
(synergistic) rather than an additive manner. Assuming independence of traits,
a scientist who is at the 90th percentile on intelligence, internal motivation,
independence, and endurance represents a one in 10,000 combination of all
these attributes.
Eysenck follows Gallon (1869, 1874) and other early researchers in
identifying high intelligence along with "zeal and industry" as primary
ingredients of great creativity. Eysenck suggests that intelligence operates
primarily through the speed with which new associations are formed. He also
 (Im)pure genius—psychoticism, intelligence, and creativity 407

Intelligence
Cognitive Knowledge
Variables Technical Skills
Special Talent

Political-religious «
Environmental Factors I CREATIVE
Variables Cultural Factors V ACHIEVEMENT
Socio-economic & I
Educational Factors i

Personality Internal motivation

Variables Confidence
Non-conformity

Figure 19.1. Creative achievement as a multiplicative function of cognitive, environmental, and

personality variables (from Eysenck, 1995, p. 39, Figure 1.4). Copyright 1995 by Cambridge
University Press. Reprinted by permission.

proposes that it is the range of associations available for problem solving that is
maximally important and that wideness of range is, in principle, independent of
speed. Thus, Eysenck suggests that intelligence and creativity are essentially
independent. In earlier work, Eysenck (1983) argued that creativity is
significantly related to IQ up to about IQ 120, but after this, becomes
independent of IQ. This has also been the view of other reviewers, none of
whom downplays the importance of intelligence (e.g., Vernon, 1987).

3.2 The role of psychoticism

Psychoticism is the active ingredient in Eysenck's theory of creativity.
Postulated as a fundamental dimension of personality, psychoticism inclines
people to all types of abnormal behaviors (see Figure 19.2, and chapter 6). Low
scorers on the psychoticism scale are characterized as high in empathy,
socialization, and co-operativeness whereas high scorers are seen as cold,
egocentric, aggressive, and tough-minded (and given to syndromes such as
psychopathy and schizophrenia). Here Eysenck follows the theory that people
who are highly original and creative differ from the vast majority in showing
behavioral quirks similar to those of schizophrenics and other psychotics.
Behavior-genetic studies suggest a common hereditary basis for great potential
and for psychopathological deviation (see chapters 6, 12, and 17).
Eysenck credits Bleuler's (1911/1978) description of the schizoid personality
for originally linking psychoticism to creativity:
He [the schizoid] is taciturn or has little regard for the effect on others of what he
says. Sometimes he appears tense and becomes irritated by senseless provocation. He
408 Further Eysenckian interests

PA

AVERAGE

iI
Figure 19.2. Psychoticism as a personality dimension. PA is the probability of a person at a given
position on abscissa developing a psychotic disorder (from Eysenck, 1995, p. 204, Figure 6.1).
Copyright 1995 by Cambridge University Press. Reprinted by permission.

appears as insincere and indirect in communication. His behavior is aloof and devoid
of human warmth; yet he does have a rich inner life. In this sense he is introverted ...
Ambivalent moods are more pronounced in the schizoid than in others, just as he
distorts the meanings of, and introduces excessive doubts into his own concepts. But
on the other hand, the schizoid is also capable of pursuing his own thoughts and of
following his own interests and drives, without giving enough consideration to other
people and to the actual realities of life. He is autistic. The better side of this autism
reveals a sturdiness of character, and inflexibility of purpose, an independence, and a
predisposition to creativity. The worst side of it becomes manifest in a lack of con-
sideration for others, unsociability, a world-alien attitude, stubbornness, ego-
centricity, and occasionally even cruelty. (Emphasis by Eysenck, 1995, p. 219)

3.3 Overinclusive thinking

But why should people with high psychoticism scores be more creative, that is,
have a wide associative horizon? Here Eysenck builds on Cameron's (1947)
and Payne, Matussek, & George's (1959) early work linking schizophrenia to
"overinclusion" in concept formation and discrimination learning. Schizo-
phrenics fail to maintain normal conceptual boundaries. Rather, they
incorporate novel elements, some of them personal, which are merely
associated and not essential into their concepts. When a child first hears a
 (Im)pure genius—psychoticism, intelligence, and creativity 409

word in a certain context, the word is associated with the entire situation (the
stimulus compound). As the word is heard again and again, only certain
aspects of the stimulus compound are reinforced. Gradually the extraneous
elements cease to evoke the response (the word), having become "inhibited"
through a lack of "reinforcement." Thus, "overinclusive thinking" may be the
result of a failure of the inhibitory process whereby learned responses like
words and concepts are circumscribed, refined, and defined. A flat associative
gradient allows the individual a wider interpretation of "relevance" as far as
responses to stimuli are concerned. This behavioral pattern has also been
described as a "looseness" of thinking or a failure to "filter out" extraneous
stimuli.

3.4 Biochemical parameters, latent inhibition and negative priming

Eysenck extends this analysis to include the biochemical studies of Gray,
Feldon, Rawlins, Hemsley, & Smith (1991), and N. S. Gray, Pickering, & J. A.
Gray (1994), as well as the experimental research on latent inhibition and
negative priming among schizophrenics and high scorers on the psychoticism
scale. He proposes that dopamine enhances creativity by reducing cognitive
inhibition, thereby overextending inclusiveness and so increasing the produc-
tion of novel combinations. Analogously, serotonin lessens creativity by
increasing inhibitory processes. It is the lack of "latent inhibition" in
suppressing remote associations that Eysenck proposes gives high psychoticism
its creative edge. Obviously psychotic thoughts differ from creative ones, so
additional cognitive characteristics including focused reasoning, general
intelligence, strong motivation, ego-strength, and the other variables listed in
Figure 19.1, come into play.
Several empirical studies have confirmed the relationship between
psychoticism and creativity. Woody and Claridge (1977) administered the
psychoticism scale from the Eysenck Personality Questionnaire and five tasks
from the Wallach-Kogan Test of Divergent Thinking (e.g., name all the things
you can think of that move on wheels; Ss responded with items such as "ball-
point pens," and "can openers") to 100 university students. Psychoticism
correlated from 0.32 to 0.45 (p < 0.05) with the "total" number of responses
produced and 0.61-0.68 for the number of "unique" responses. No reliable
correlations were found between creativity and extraversion and neuroticism,
but the lie-score, which correlates negatively with psychoticism and is partly a
measure of social conformity, showed consistent negative correlations with the
creativity scores.
In a study of Canadian university students, Rushton (1990) replicated
Woody and Claridge's (1977) correlation between psychoticism and creativity.
Rushton also showed that creativity scores correlated with IQ. Using real-life
criteria, 337 professional artists with a record of holding successful exhibits,
410 Further Eysenckian interests

were administered the Eysenck Personality Questionnaire and found to have

higher scores on the psychoticism scale than nonartists (K. O. Gotz & K. Gotz,
1979a; 1979b).

4. PSYCHOTICISM AND "THE MAD SCIENTIST"

4.1 Review of the literature
The portrayal of the "scientific personality" in some biographies leaves little
doubt as to what characterizes the ideal scientist: objectivity, emotional
neutrality, rationality, open-mindedness, superior intelligence, integrity, and a
communal, open, and co-operative attitude toward sharing knowledge. Indeed,
sometimes the history of science is "as inspiring in its human values as are the
legends of the Saints" (Knickerbocker, 1927, p. 305).
Eysenck provides many biographical vignettes that document the gap
between reality and this idealized portrait. Scientists often engage in
emotionally charged ideological battles, where personal success and the
destruction of opponents are more important than objectivity, where selective
perceptions and distortion of facts qualify the notion of rationality, and where
personal biases lead to editorial rejection of contrary ideas. Outright deception
and fraud mar the ideal of honest integrity, and secrecy, suspicion, and
aggressive competition in the race to be "number one" are as manifest as any
altruistic desire to share knowledge and cooperate. Nonetheless, some
inspirational qualities do come through, as in Eysenck's examples of "un-
conquerable will" and achievement (as exemplified by George Washington
Carver who was born a slave but rose against the odds to considerable heights
of scientific achievement—see chapter 15).
The investigation into the psychological characteristics of eminent scientists
began with Francis Gallon (1869,1874). His pioneering work was expanded by
Cattell (1903, 1910), Havelock Ellis (1904), Cox (1926), Roe (1952), Cattell
and Drevdahl (1955), Terman (1955), and by Taylor and Barron (1962), and
others (see Jackson & Rushton, 1987, and Sulloway, 1996, for reviews). From
this growing body of research it became clear that successful scientists are not
at all "Saint-like" in either their personality or work style. They often display
reclusive personalities, arrogant work styles, hostile responses to frustration,
and intrinsic motivations bordering on autism.
For instance, Terman's (1955) longitudinal study of 800 high-IQ men found
that those who took science degrees at college differed from nonscientists in
showing great intellectual curiosity from an early age and in being lower in
sociability than average. Terman concluded that "the bulk of scientific research
is carried on by devotees of science for whom research is their life and social
relations are comparatively unimportant" (p. 7). Cited is the work of Roe
(1952), who found scientists to have difficulty in interpersonal situations and to
often try to avoid them. Terman described Roe's sample of scientists as tending
 (Im)pure genius—psychoticism, intelligence, and creativity 411

"to be shy, lonely, slow in social development, and indifferent to close personal
relationships, group activities, or politics" (p. 7; see chapter 20 for details).
Terman noted that such traits were not necessarily defects of personality, for
emotional breakdowns were no more common than among nonscientists.
Instead, he suggested that a below-average interest in social relations and a
heavy concentration of interest in the objective world was a normal departure
from average that was decidedly favorable for the professional development of
a scientist.
Cattell's (1962, 1965) and Cattell and Drevdahl's (1955) profile of the
prototypic scientist emerges from both the qualitative study of biographies and
from quantitative psychometric studies of leading physicists, biologists, and
psychologists. Cattell found successful scientists to be reserved and introverted,
intelligent, emotionally stable, dominant, serious-minded, expedient, venture-
some, sensitive, radically thinking, self-sufficient, and having a strong and
exacting self-concept. He noted that the physicists, biologists, and psychologists
were similar in personality except that psychologists were less serious-minded
and more "surgent" and talkative than nonpsychologists. Creative scientists
differed most from normals on schizothymia-cyclothymia factor, with scientific
researchers being toward the schizothymic end. Cattell thus describes scientists
as being skeptical, internally preoccupied, precise, and critical individuals who
are exacting and reliable.
Several studies were carried out by Barren and his colleagues (Barron, 1962;
Taylor & Barron, 1962). Barron, for example, found creative people generally
to be cognitively complex (preferring complexity and imbalance in pheno-
mena), to have a more differentiated personality structure, to be independent
in their judgment and less conformist in social contexts such as the Asch group
pressure situation, to be self-assertive and dominant, and to be low in using
suppression as a mechanism for the control of impulses and thoughts (that is,
they forbade themselves fewer thoughts). Chambers (1964) compared eminent
researchers with those not so eminent but matched on other relevant variables.
Results indicated that the more creative scientists were also more dominant,
had more initiative, were more self-sufficient, and were more motivated toward
intellectual success. McClelland (1962) found successful scientists to be not
only higher in need for achievement but also to be calculating risk-takers in the
same way as are successful business entrepreneurs. The risk-taking, however,
involved dealing with nature or physical situations rather than social situations,
for he, too, found scientists to be decidedly avoidant of interpersonal
relationships. Scientists, for instance, indicated a much higher preference for
being a lighthouse keeper as opposed to being a headwaiter (Item no. 324 on
the Strong Vocational Interest Blank). McClelland also argued that the need
for scientific achievement was a strong aggressive drive "which is normally kept
carefully in check and diverted into taking nature apart" (1962, p. 162). In
short, the scientist is "introverted and bold" (Drevdahl & Cattell, 1958).
412 Further Eysenckian interests

Studies of psychologists have found that publication and citation counts can
be predicted by those components of achievement motivation that concern the
enjoyment of challenging tasks and hard work, but not by those components
concerned with interpersonal competition or bettering others (Helmreich,
Beane, Lucker, & Spence, 1978; Helmreich, Spence, Beane, Lucker, &
Matthews, 1980). Type A "workaholic" behavior (aggressive, incessantly
struggling, time-oriented, hostile when frustrated) predicts the number of
citations a psychologist's work earned from others (Matthews, Helmreich,
Beane, & Lucker, 1980). Using structural equation modeling, Feist (1993)
found a good fit leading from hostile personality, internal motivation, and
arrogant working style to objectively measured eminence in 100 physicists,
chemists, and biologists at major research universities in California.
Rushton, Murray, and Paunonen (1983) examined the relation of 29
personality traits to research and teaching effectiveness composites (which
intercorrelated zero) in two independent samples of Canadian university
professors. The cluster of traits associated with being an effective researcher
differed from those characterizing the effective teacher. As indicated in Figure
19.3, the attributes of the successful researcher were less socially desirable than
those associated with being a good teacher (ambitious, enduring, dominant,
aggressive, independent, and demanding definiteness versus liberal, sociable,
extraverted, and supportive; indeed, good researchers were characterized as
less "objective" than good teachers and more defensive and authoritarian).
The only variables loading positively on both dimensions were intelligence and
leadership, while meekness suggested being poor in both. Although this study
was not carried out to test Eysenck's theory, it seems that successful
researchers are high on high psychoticism characteristics.

4.2 New evidence

To test whether the profile of the successful researcher in Rushton et al.'s
(1983) study did conform to high psychoticism, I sought Professor Eysenck's
help in weighting each of the 29 traits used (see Table 19.2) from -3 (strong
negative correlation with psychoticism) to +3 (strong positive correlation with
psychoticism).
Although I published this follow-up study in Personality and Individual
Differences (Rushton, 1990) and although the results confirmed Eysenck's
theory of creativity, the study was inexplicably overlooked in the literature
reviewed for the Genius book! This therefore provides me with an especially
good justification for re-reporting those data again here!
 (Im)pure genius—psychoticism, intelligence, and creativity 413

90

RESEARCH*
70 •AMBITIOUS

• ENDURING
SEEKS DEFINITENESS* ^

50 •DOMINANT

•LEADERSHIP
AGGRESSIVE* 40 •INTELLIGENT
INDEPENDENT*
30
DEFENSIVE*
2O'
AUTHORITARIAN*
10
ANXIOUS-
90 -8b -7& -60 -50 -4O -30 -20 -10 10 20 30 40 50 60 70*T£fiCH\i
-10 9L8ERAL

OBJECTIVE. •EXTRAVERTED
-20
AESTHETICALLY- •SOCIABLE
-30 SENSITIVE •SUPPORTING
MEEK*
-40

-50'

-60

-70

-80

-90

Figure 19.3. Plot of mean factor pattern coefficients of personality traits on dimensions of research
productivity and teaching effectiveness, averaged across two studies. Only those trails with absolute
values of >.30 on either factor in both studies are shown. (Based on data in Rushton et al., 1983.)

5. PSYCHOTIC PROFESSORS?

In the first of the two studies, the participants were 46 male and six female full-
time psychology professors at the University of Western Ontario. (Due to the
small number of females, all analyses are collapsed across sex.) Each professor
was assessed on 29 traits using four techniques: faculty-peer ratings, student-
ratings, self-ratings, and self-report-questionnaires (Rushton et al., 1983).
Ratings were made on nine-point scales, using the trait names and brief
descriptions shown in Table 19.2, which also shows the split-half reliabilities for
the faculty-peer and student judgments. Instructions emphasized that ratings
Table 192. Split-half reliabilities of peer and student ratings of personality computed across
Professor targets for each of 29 personality traits (decimals omitted). Also shown is the weighting
assigned to the trait for its loading on psychoticism (P)

Raters
Personality trait and trait definition Faculty Students P
(n = 52) (n = 43) Weighting

1. Meek (mild mannered; subservient) 73 57 -3

2. Ambitious (aspiring to accomplish difficult tasks; 88 74 +1
striving, competitive)
3. Sociable (friendly, outgoing, enjoys being with people) 74 63 -2
4. Aggressive (argumentative, threatening; enjoys 84 62 +3
combat and argument)
5. Independent (avoids restraints; enjoys being 80 42 +2
unattached)
6. Changeable (flexible, restless; likes new and different 77 33 0
experiences)
7. Seeks definiteness (dislikes ambiguity or uncertainty 84 22ns +1
in information; wants all questions answered
completely)
8. Defensive (suspicious, guarded, touchy) 72 56 +3
9. Dominant (attempts to control environment; forceful, 87 60 +2
decisive)
10. Enduring (willing to work long hours; persevering, 90 52 0
steadfast, unrelenting)
11. Attention seeking (enjoys being conspicuous, 88 67 +1
dramatic, colorful)
12. Harm avoiding (careful, cautious, pain-avoidant) 84 90 -2
13. Impulsive (spontaneous, hasty, impetuous, and 89 31 +3
uninhibited)
14. Supporting (gives sympathy and comfort; helpful, 84 36 -3
indulgent)
15. Orderly (neat and organized; dislikes clutter, 77 56 -1
confusion, lack of organization)
16. Fun loving (playful, easygoing, light-hearted; does 88 75 0
many things "just for fun")
17. Aesthetically sensitive (sensitive to sounds, sights, 80 74 0
tastes, smells)
18. Approval seeking (desires to be held in high esteem; 76 42 -2
obliging, agreeable)
19. Seeks help and advice (desires and needs support, 80 86 -2
protection, love, advice)
20. Intellectually curious (seeks understanding; 78 65 0
reflective, intellectual)
21. Anxious (tense, nervous, uneasy) 60 63 0
22. Intelligent (bright, quick, clever) 89 50 0
23. Liberal (progressive, seeks change, modern, 81 29ns 0
adaptable)
24. Shows leadership (takes initiative and responsibility 86 54 0
for getting things done)
25. Objective (just, fair, free of bias) 78 48 0
26. Compulsive (meticulous, perfectionistic, concerned 69 50 0
with details)
27. Authoritarian (rigid, inflexible, dogmatic, 70 52 0
opinionated)
28. Extraverted (has many friends; craves excitement; 90 71 0
fond of practical jokes; is carefree, easygoing, optimistic)
29. Neurotic (a worrier; overly emotional; anxious, 61 71 0
moody, and often depressed)
Mean 79 56
Note. From Rushton (1990, p. 1079, Table 5). Copyright 1995 by Elsevier Science Ltd. Reprinted
by permission.
 (Im)pure genius—psychoticism, intelligence, and creativity 415

were to be made relative to other professors rather than to people in general.

There was an average of 12 ratings per faculty member.
The various personality assessments showed convergent validity with the
scores on self-ratings and questionnaires averaging 0.52 across the 29 traits and
the ratings made by faculty peers and by students averaging 0.43. Because the
return rates for the peer-ratings (n = 52) were higher than for other
procedures, the analyses will be limited to these. The ratings were combined
using Professor Eysenck's assigned weights (shown in Table 19.2) to produce a
psychoticism score.
An index of creativity was made from two measures of research
effectiveness: (1) total publications over the previous four years as listed in
either the Social Science Citation Index or the Science Citation Index
(whichever was larger for the particular individual and with credit assigned
equally for senior and junior authorship); and (2) total citations for the
previous three years in the same Citation Indices (with first authored self-
citations excluded). Year-to-year stability was 0.60 for publications and 0.98 for
citations. The two indices intercorrelated 0.28 (p < 0.05) and were combined
(using averaged standard scores). The correlation between psychoticism and
creativity was 0.40 (p < 0.01).
A follow-up study using a mail survey was made at nine other psychology
departments in Canada with 69 (68 male, one female) people responding
(Rushton et al., 1983). The same 29 personality traits and definitions as in
study 1 were used. Respondents were instructed to rate themselves in
percentiles, "relative to other Canadian university psychology professors." The
distributions turned out to be roughly normal, with a mean percentile across
traits of 55 and a standard deviation of 21. Socially desirable traits were rated
higher than socially undesirable traits, with professors rating themselves at the
80th percentile on intelligence and at the 26th percentile on authoritarianism!
Four items were aggregated to index creativity: (1) total number of
publications; (2) mean number of publications in last five years; (3) number
of hours spent on research; and (4) rated enjoyment of research. Each of these
was significantly associated with the others (mean correlation of 0.36; p <
0.01). Psychoticism correlated with creativity 0.43 (p < 0.01).

6. ACTS OF DESTRUCTION

The two studies presented of Canadian university professors confirm Eysenck's

predicted relationship between psychoticism and creativity although they were
silent about the mechanisms involved. However, personality traits must exert
their effects either through the cognitive system or through the social system.
As reviewed, Eysenck proposes that psychoticism works principally by
widening an individual's associative network available for problem solving.
416 Further Eysenckian interests

Eysenck's (1995) psychology of creativity will likely go on to become another

"citation classic." To this reviewer, it is difficult to find criticism to offer, for
Eysenck seems so very right about so very much! Rather than criticism,
therefore, I will highlight some topics that struck me as in some ways
paradoxical and so perhaps especially worthy of future study—the role of
intelligence, motivation, values, and the social management of research teams,
bureaucracies, and public relations.
Several reviewers have concluded that creativity is significantly related to IQ
up to about IQ 120 (the level of an average North American university
undergraduate). Beyond IQ 120, creativity becomes independent of IQ
(Eysenck, 1983; Vernon, 1987). Because little evidence is provided for this
claim, it may be premature. Individuals with IQs of 120 would have great
difficulty competing successfully in some of today's most creative scientific
professions (astrophysics, computer engineering, mathematics). Moreover, the
importance of general cognitive ability has now been shown in literally
hundreds of studies to predict work performance in all occupations, whether
measured by supervisor ratings, training success, job knowledge, work sample,
or ongoing job learning, with validities as high as r = 0.80 (see Herrnstein &
Murray, 1994, for review). Many of these studies were carried out on very large
samples by the U.S. Employment Service and the U.S. Armed Services
examining jobs rated as of low, medium, and high complexity, or categorized as
clerical, professional, or technical. Meta-analyses showed that general
cognitive ability, rather than specific cognitive aptitudes or job knowledge,
was the best predictor of performance in all cases. Typically, as the complexity
of the job increases, the better cognitive ability predicted performance (e.g.,
managers and professions 0.42-0.87, sales clerks and vehicle operators 0.27-
0.37; e.g., Hunter, 1986, Table 1).
Arousal and motivation constitute another topic where evidence is
conflicting (Eysenck, 1995, pp. 267-270). Based on biographies, low arousal
and even dream-like reveries are often associated with many acts of creation;
and experimental studies dovetail by showing that high arousal narrows
attention. Yet creative people are also often anxious and introverted and so are
more aroused than average. Eysenck concludes that perhaps the supreme act
of creation occurs during low arousal—and that high arousal accompanies the
elaboration stage, when creative people attempt to prove their intuitive
insights, argue with skeptics, and so on.
Eysenck's distinction makes sense, but new attention might be given to the
obsessive-compulsive and highly aroused nature of much creativity. As Jensen
(1996) has pointed out, the ordinary term "motivation" explains little and
seems too intentional and self-willed to fit the behavior of geniuses whose
biographies show that although they may occasionally have to force themselves
to work, they cannot will themselves to be obsessed by the subject of their
work. Their obsessive-compulsive mental activity in a particular sphere is
 (Im)pure genius—psychoticism, intelligence, and creativity 417

virtually beyond conscious control. Instead, Jensen likens "mental energy" in

geniuses to the kind of cortical arousal seen under the influence of stimulant
drugs. Jensen elaborates on a clue offered by Havelock Ellis (1904) that
eminent men suffered from gout—a painful inflammation, usually of the joint
in the big toe, caused by the formation of uric acid crystals. Dozens of studies
(reviewed by Jensen & Sinha, 1993) show that although serum urate level
(SUL) is slightly correlated with intelligence, it is much more highly correlated
with achievement and productivity. For instance, among high school students
there is a relation between scholastic achievement and serum urate level, even
after controlling for IQ. Another study found a correlation of 0.37 between
serum urate level and the publication rates of university professors (Mueller &
French, 1974). One explanation is that the molecular structure of uric acid is
similar to caffeine and therefore acts as a brain stimulant. This energy,
combined with very high intelligence or an exceptional talent, results in high
productivity.
Another aspect of the motivational structure of geniuses identified by Jensen
from biographical material is a powerful value system that channels and
focuses the individual's mental energy. It is not something mundane, but seems
to control the direction of personal ambition and the persistence of effort and
also seems at odds with some aspects of the psychopathic character implied
earlier. People are often puzzled by what they perceive as the genius's self-
sacrifice of his other needs (as well as to his often egocentric indifference to the
needs of others). But the genius's value system, at the core of the self-concept,
is hardly ever sacrificed for the typical pleasures of ordinary persons. Acting on
their own values—perhaps one should say acting-out their self image—is a
notable feature of famous geniuses.
A less surprising proclivity, often manifested at an early age, is unusually
strong and long-lasting curiosity and exploratory behavior. Charles Darwin
himself stated in his autobiography (Barlow, 1958 p. 141) that he had always
had the strongest desire to understand and explain whatever he observed, that
is, to group all facts under some general laws. A perhaps related "superego"
trait is a concern for excellence, and especially for "elegant, virtuous, and
beautiful solutions." [I am indebted to Jackson (1987) for some of the
discussion in this and the next section.] Achieving "virtuous" solutions (Robert
Oppenheimer's poetic phrase) requires long hours of arduous work mastering
complex and sometimes recalcitrant problems. Searching for beauty and virtue
is a quite different view of research than that taken by the typical corporate
financial officer or university administrator, who see research as a means to the
end of either profits or enhanced institutional prestige.
To summarize Eysenck (1995), Jensen (1997), and others on great scientists:
Genius = Ability x Productivity x Creativity x Values x Curiosity
418 Further Eysencldan interests

where Ability = intelligence and information processing efficiency; Productiv-

ity = endogenous cortical stimulation; Creativity = trait psychoticism; Values
= central motivating mechanisms for honest, beautiful, and virtuous solutions;
and Curiosity = the search for general laws. Even the above list of synergistic
traits is not exhaustive.
Managing the social world of science is a much neglected topic but obviously
calls for another set of traits, especially in the increasingly complicated, high-
tech, bureaucratized world of "Big Science." Major innovations need to be
"sold" through networking and social organizations, including government
bodies, the mass media, funding agencies, and scientific and professional
groups. Perhaps high-psychoticism scorers are less vulnerable to social
blandishments and criticisms. But it would be a mistake to fixate on maverick
and hostile Don Quixotes tilting at establishment windmills. Because a fairly
clear personality profile of the innovative researcher emerges, it does not mean
that individual differences do not exist! For example, as Cattell (1962) noted,
although many scientists have historically been recognized as less sociable than
average, Galileo, Leibnitz, and others were as fully at home in the social free-
for-all of court circles as in the laboratory. Far from thinking that the optimal
condition for scientific advance is absolute freedom to think in a far removed-
from-it-all environment, evidence (e.g., Pelf, 1967; reviewed in Jackson, 1987)
suggests that social constraints, and constructive social interaction are
beneficial.
Many of the more exciting episodes in science (some discussed in Eysenck)
occur when a scientist lives to see himself overthrowing (or at least going well
beyond) established thinking. Here Charles Darwin again comes immediately
to mind. The converse is similarly fascinating, as when the establishment
overthrows the scientist! Here a good example is Ignaz Philipp Semmelweis
who failed to persuade the medical establishment of Hungary that washing
hands in hospitals saved lives (Eysenck, 1995, pp. 150-153). Darwin's brilliant
orchestration of his output and ability to attract loyal and powerful advocates
contrasts with Semmelweis's intemperate, self-destructive, and counter-
productive behavior. The role of marshaling public relations to win scientific
battles against great odds needs much further investigation. [A recent book by
Duesberg (1996) is a masterpiece of describing how the majority of scientists
who believed that AIDS is caused by a virus energized the entire scientific
community behind them leaving heretics like Duesberg out in the cold.]
Personally I am intrigued by the "antiscientific" revolution led by Franz Boas
who almost single-handedly succeeded in decoupling social science from
Darwinian thinking (cf. Rushton, 1995). Boas must have been a public-
relations genius... especially given that so much of what he had to say turns out
to be so false!
 (Im)pure genius—psychoticism, intelligence, and creativity 419

Genius, including evil genius (your pick) clearly needs more study! Acts of
genius are acts of creative destruction (destroying old theories, time-honored
beliefs, and prejudices). We need to learn how to foster creativity and live with
the concomitant destruction.

REFERENCES

Barlow, N. (Ed.) (1958). The autobiography of Charles Darwin. New York: W. W.

Norton.
Barren, F. (1962). The needs for order and disorder as motives in creative activity. In C.
W. Taylor, & F. Barren (Eds.), Scientific creativity: Its recognition and development
(pp. 139-152). New York: Wiley.
Bleuler, M. (1911, translation 1978). The schizophrenic disorders. New Haven: Yale
University Press.
Cameron, N. (1947). The psychology of behavior disorders. Boston: Houghton Mifflin.
Cattell, J. M. (1903). A statistical study of eminent men. Popular Science Monthly, 32,
359-377.
Cattell, J. M. (1910). A further statistical study of American men of science. Science, 32,
633-648.
Cattell, R. B. (1962). The personality and motivation of the researcher from
measurements of contemporaries and from biography. In C. W. Taylor & F. Barren
(Eds.), Scientific creativity: Its recognition and development. New York: Wiley.
Cattell, R. B. (1965). The scientific study of personality. Harmondsworth, U.K.: Penguin.
Cattell, R. B., & Drevdahl, J. E. (1955). A comparison of the personality profile (16 PF)
of eminent researchers with that of eminent teachers and administrators, and of the
general population. British Journal of Psychology, 46, 248-261.
Chambers, J. A. (1964). Creative scientists of today. Science, 145, 1203-1205.
Cox, C. M. (1926). The early mental traits of three hundred geniuses. Stanford, CA:
Stanford University Press.
Dennis, W. (1955). Variation in productivity among creative workers. Scientific Monthly,
80, 277-278.
Drevdahl, J. E. & Cattell, R. B. (1958). Personality and creativity in artists and writers.
Journal of Clinical Psychology, 14, 107-111.
Duesberg, P. (1996). Inventing the AIDS virus. Washington, DC: Regnery.
Ellis, H. (1904). A study of British genius. London: Hurst and Blackett.
Endler, N. S., Rushton, J. P., & Roediger, H. L. (1978). Productivity and scholarly
impact (citations) of British, Canadian, and U.S. departments of psychology.
American Psychologist, 33, 1064-1083.
Eysenck, H. J. (1983). The roots of creativity: Cognitive ability or personality trait?
Roeper Review, 5, 10-12.
Eysenck, H. J. (1995). Genius: The natural history of creativity. Cambridge: Cambridge
University Press.
Feist, G. J. (1993). A structural model of scientific eminence. Psychological Science, 4,
366-371.
420 Further Eysenckian interests

Gallon, F. (1869). Hereditary genius: An inquiry into its laws and consequences. London:
Macmillan.
Gallon, F. (1874). English men of science. London: Macmillian.
Golz, K. O., & Golz, K. (1979a). Personality characlerislics of professional artisls.
Perceptual and Motor Skills, 49, 327-334.
Golz, K. O., & Golz, K. (1979b). Personality characlerislics of professional artisls.
Perceptual and Motor Skills, 49, 919-924.
Gray, J. A., Feldon, J., Rawlins, J. P., Hemsley, D. R., & Smith, A. D. (1991). The
neuropsychology of schizophrenia. Behavioral and Brain Sciences, 14, 1-84.
Gray, N. S., Pickering, A. D., & Gray, J. A. (1994). Psycholicism and dopamine D2
binding in Ihe basal ganglia using single photon emission tomography. Personality and
Individual Differences, 17, 431-434.
Helmreich, R. L., Beane, W. E., Lucker, G. W., & Spence, J. T. (1978). Achievemenl
motivation and scientific attainment. Personality and Social Psychology Bulletin, 4,
222-226.
Helmreich, R. L., Spence, J. T., Beane, W. E., Lucker, G. W., & Matthews, K. A.
(1980). Making il in academic psychology: Demographic and personality correlates of
attainment Journal of Personality and Social Psychology, 39, 896-908.
Herrnslein, R. J., & Murray, C. (1994). The bell curve. New York: Free Press.
Hunter, J. E. (1986). Cognitive ability, cognitive aptitudes, job knowledge, and job
performance. Journal of Vocational Behavior, 29, 340-362.
Jackson, D. N. (1987). Scientific and technological innovation: Ils personological and
motivational context In D. N. Jackson and J. P. Rushlon (Eds), Scientific excellence:
Origins and assessment. Beverley Hills, CA: Sage.
Jackson, D. N., & Rushton, J. P. (Eds.) (1987). Scientific excellence: Origins and
assessment. Beverly Hills, CA: Sage.
Jensen, A. R. (1996). Gifledness and genius: Crucial differences. In C.P. Benbow & D.
Lubinski (Eds.), Intellectual talent: Psychometric and social issues (pp. 393-411).
Baltimore: The Johns Hopkins University Press.
Jensen, A. R. & Sinha, S. N. (1993). Physical correlates of human intelligence. In P. A.
Vernon (Ed.), Biological approaches to the study of human intelligence (pp. 139- 242).
Norwood, NJ: Ablex.
Knickerbocker, W. S. (1927). Classics of modem science. New York: Knopf.
McClelland, D. C. (1962). The calculated risk: An aspect of scientific performance. In
C. W. Taylor & F. Barren (Eds.), Scientific creativity: Its recognition and development.
New York: Wiley.
Matthews, K. A., Helmreich, R. L., Beane, W. E., & Lucker, G. W. (1980). Pattern A,
achievemenl slriving, and scientific merit: Does pattern A help or hinder? Journal of
Personality and Social Psychology, 39, 962-967.
Mueller, E. F. & French, J. R., Jr. (1974). Uric acid and achievemenl. Journal of
Personality and Social Psychology, 30, 336-340.
Payne, R. W., Malussek, P., & George, E. I. (1959). An experimenlal study of
schizophrenic Ihoughl disorder. Journal of Mental Science, 105, 627-652.
Pelf, D. C. (1967). Creative tensions in Ihe research and developmenl climate. Science,
157, 160-165.
Roe, A. (1952). The making of a scientist. Weslport, CT: Greenwood Press.
 (Im)pure genius—psychoticism, intelligence, and creativity 421

Rushton, J. P. (1989). A 10-year scientometric revisit of British psychology departments.

The Psychologist, 2, 64-68.
Rushton, J. P. (1990). Creativity, intelligence, and psychoticism. Personality and
Individual Differences, 11, 1291-1298.
Rushton, J. P. (1995). Race, evolution, and behavior. New Brunswick, NJ: Transaction.
Rushton, J. P., Murray, H. G., & Paunonen, S. V. (1983). Personality, research
creativity, and teaching effectiveness in university professors. Scientometrics, 5, 93-
116.
Science Citation Index. Philadelphia, PA: Institute for Scientific Information.
Shockley, W. (1957). On the statistics of individual variations of productivity in research
laboratories. Proceedings of the Institute of Radio Engineers, 45, 279-290.
Simonton, D. K. (1988). Scientific genius: A psychology of science. New York: Cambridge
University Press.
Social Science Citation Index. Philadelphia, PA: Institute for Scientific Information.
Sulloway, F. J. (1996). Bom to rebel. New York: Pantheon.
Taylor, C. W. & Barron, F. (1962). Scientific creativity: Its recognition and development.
New York: Wiley.
Terman, L. M. (1955). Are scientists different? Scientific American, 192, 25-29.
Vernon, P. E. (1987). Historical overview of research on scientific abilities. In D. N.
Jackson & J. P. Rushton (Eds.), Scientific excellence: Origins and assessment (pp. 40-
66). Beverly Hills, CA: Sage.
Walberg, H., Strykowski, B. F., Rovai, E., & Hung, S. S. (1984). Exceptional
performance. Review of Educational Research, 54, 87-112.
Woody, E. & Claridge, G. (1977). Psychoticism and thinking. British Journal of Social
and Clinical Psychology, 16, 241-248.
Chapter 20
Molecular creativity, genius and madness

H. Nyborg

1. INTRODUCTION

Creativity has always been important for the survival of individuals, the group
and society in general, and its importance is likely to grow with time. Energy
becomes more and more concentrated in chemical and physical systems.
Political, executive, and military powers can be canalized electronically by a
light touch of a button in an increasingly complex high-tech world. Just one
unimaginative political, industrial, or military leader may, under the worst
possible conditions, cause havoc, where the consequences of similar acts would
previously have taken a more local and less damaging form. Large modern
technical corporations now go broke by a few unwise decisions, and others
prosper with a single stroke of genius and enjoy previously unheard-of profit.
Researchers are now pressed to identify creativity and teachers to cultivate it.
The launching of the first Soviet sputnik illustrates the point. Shortly after this
event there was a sudden and hectic interest in the U.S.A. to find better ways to
promote creativity and put the nation back in its leading role. Systematic
comparisons among nations began to appear, leading nations found themselves
lagging behind, and everybody agreed that something had to be done. Where
are we today?
Over time we see many attempts to define, measure, and promote
creativity—to mention but a few, Albert (1992), Amabile (1983), Cattell
(1903), Cattell and Drevdahl (1955), Cox (1926), Glover, Ronning, and
Reynolds (1989), M. Goertzel, V. Goertzel, and T. Goertzel (1978), V.
Goertzel and M. G. Goertzel (1962), Guilford (1950), Jackson and Rushton
(1987), Kasparson (1978), Lombroso (1901), MacKinnon (1961), Mansfield
and Busse (1981), Ochse (1990), Radford (1990), Roe (1953), Sternberg
(1985), Terman (1925), Terman and Oden (1947), Vernon (1982), and
Zuckennan (1977).
What is the general lesson emanating from this impressive amount of
research? Do we see fair agreement about the precise and empirical definition
of creativity? Have we achieved a useful identification of relevant causes of
 Molecular creativity, genius and madness 423

creativity and do we see transparent operationalization of mediating

mechanisms? Do researchers agree on the proper level for analyzing
creativity, or on the number of analytic dimensions needed to describe it?
Do explanations demonstrate unity and coherence? Have the educational
systems finally been handed powerful tools for cultivating creativity or, at least,
been provided with the means for not standing in the way of promising
creativity or seeing itself brutally suppressing it?
I am afraid the answer is: No. Neither can any of the questions can be
answered in the affirmative, nor am I the only pessimist. The recent Handbook
of Creativity by Glover et al. (1989) came to the extraordinary conclusion that
creativity research seems to degenerate. There is something terribly wrong
here: Creativity increases in importance in modern societies, but research on it
stalls. One possible reaction to this situation is to try and redefine the problem
and/or to find new ways of studying it.
Both paths are chosen in this chapter. Obviously, such a radical move
involves heavy risks and little promise of success. But what is there to loose?
The chapter comes in three sections. The first section briefly discusses major
obstacles in contemporary creativity research. Section two presents Hans
Eysenck's way of attacking them. The last section redefines creativity, genius,
and madness in purely physico-chemical terms, and suggests an entirely
molecular natural science approach to study them (Nyborg, 1994).

2. OBSTACLES TO CURRENT RESEARCH ON CREATIVITY

Extraordinary creativity often pops up quite unexpectedly in a family, and

there are surprisingly few records of families producing two pure geniuses in
succession. Genius sometimes blossoms in poverty-ridden places with no good
schools, or surfaces in homes entirely lacking in academic tradition. There are
moving stories of how extremely creative individuals survived despite growing
up in the prohibitive shadow of terrified or disappointed parents, not
understanding a word spoken by their strange child, or in strong opposition
to orthodox teachers or a church that knows better. The unfolding of genius in
such places or under such circumstances remains a complete mystery to
contemporary research on creativity, and challenges theories of creativity
based on socialization or role-modeling. The many accounts of outrageous
repression of true genius illustrate some further points. Extraordinary
creativity is often seen neither as a blessing to the person in question, nor to
the immediate surroundings. Sometimes no-one profits from it for a long time
or ever. The many fascinating life-history accounts of geniuses rarely reveal
anything of scientific interest with regard to the springs of extraordinary
creativity. Both favorable and oppressive circumstances may befit the unfolding
of true genius. It is unfortunate, that while detailed life-history accounts
424 Further Eysenctdan interests

certainly illustrate the circumstances (favorable or not) under which genius

unfolds, they typically reveal nothing about how the genius came to his
extraordinary capacity.
The genetic aspect of genius is also essentially untrodden land. Gallon's
(1869/1978) study led him to believe that inheritance explained much of the
individual variability in genius. This conclusion is beset with problems. In fact,
Gallon studied acknowledged excellence in various areas rather lhan genius in
any precise sense. Worse, Eysenck (1995) found lhal Gallon aclually turned
Ihe excellence-inherilance argumenl on ils head, so lhal his resulls are
probably better explained in lerms of Ihe environment Sludies specifically
designed lo reveal Ihe genelic basis for crealivity, such as twin studies, give
disappointing resulls. Nichols (1978) found a modesl 20% genelic influence,
and when Canler (1973) controlled for IQ, the genelic impact shrank lo
nolhing. The few designated genelic studies ihus show weak, or no familial
Iransmission of creativity. They neilher lell us where extremely high creativity
comes from, nor why a genius rarely, if ever, leaves offspring of equal standing.
The discussion of "emergenesis" by Lykken (1982) and Lykken, McGue,
Tellegen, & Bouchard (1992) mighl Ihrow new lighl on why creativity suddenly
appears in a family and slill may have a significant genelic componenl, bul
righl now we simply have loo few genetic dala lo say anything of scientific
importance on Ihe mailer. The sludy of a common genelic basis for familial
aggregation of creativity and psychopalhology may hold more promises (see
chapters 6, 17, and 19).
Socialization Iheory fares no belter lhan genelic Iheory in explaining Ihe
facls. There are many books and courses guaranteeing quick and easy progress
in Ihe promotion of personal, occupational, scholastic, scientific, or artistic
creativity. None of Ihe promises translate into verifiable generalized effecls on
crealive ability or achievemenl, however. Children as well as adulls can be
laughl how lo lake creativity lesls, and Ihis certainly raises iheir crealivity
score, bul Ihe training seems nol lo generalize lo anything useful outside the
realm of taking Ihe test, quite like training intelligence.
To sum up, high creativity can neilher be explained by anecdolal evidence,
nor by contemporary genelic, social, or pedagogical Iheory.
In real science such a confusing state in an important area would have
attracted an army of scientists equipped with a healthy taste for stringenl basic
research. They would eagerly try to map verifiable causes and mechanisms,
bring creativity under lighl experimental conlrol and Ihen develop proper
means for furthering it. Apparently, Ihis is nol whal happens in traditional
psychology and sociology of creativity. Some researchers continue lo study
biographies or talk lo crealive people, lo see how Ihey Ihemselves Ihink they do
Iheir unusual Iricks. Others study Ihe childhood of precocious children or
adulls, or try lo establish connections between creativity, personality, ability,
and Ihe environmenl by questionnaires and factor analyses. Slill olhers Irain
 Molecular creativity, genius and madness 425

people in "alternative" or "flexible" thinking (often for good money), but

rarely follow up on the decisive question of whether the training generalized to
creative acts in related or different contexts. Recently, it has become fashion-
able to put the cart before the horse, that is, to first develop complex theories
of multiple intelligence and creativity, and then to defend or excuse these
theories in the absence of hard experimental evidence and solid predictions.
The acid test of any approach to creativity is, of course, to skip the grand
visions and begin asking pertinent questions about the psychometric
properties, the quality of the experimental control, the predictive value, and
the pedagogical effect of the work. However, contemporary research does not
even offer a clear answer to simple questions like: "Can you, on basis of your
theory or practice, tell me how to train the creativity of this particular child or
adult, and can you guarantee that the treatment has lasting and generalizable
effects?"; "Can you, in retrospect, identify the specific factors that, given your
theory, explain beyond reasonable doubt why this particular person became
extraordinarily creative, and the other person not?" As mentioned, families
with aggregation for psychopathology also tend to show aggregation for
creativity. However, the notion of common or codominant genes begs the
question: Which genes? To say that creativity or genius is a product of society,
active learning or role-modeling is to beg the question of why controlled
socialization and specific learning experiments utterly fail in explaining or
promoting anything creative. Glover et al. (1989) seem indeed justified in their
critical view of much contemporary creativity research. Instead of leading to
new glorious heights, it basically produces more anecdotes, more untestable
theories, and more excuses for predictive impotence.

3. HANS EYSENCK'S APPROACH TO CREATIVITY

3.1 Introduction
Eysenck entered the area of creativity research in the early 1980s, but he
certainly did not start from scratch. Many years earlier he contemplated how
creativity squares with intelligence. With his sound habit of posing simple
questions in complicated matters, he first inquired into whether creativity is a
cognitive ability or a personality trait (Eysenck, 1983). He also wondered
whether creativity squares with S. Eysenck's and his own psychoticism (P)
dimension (H. J. Eysenck & S. B. G. Eysenck, 1975; Eysenck, 1989,1993a). In
the early 1990s (Eysenck, 1993b), he was finally prepared to sketch a new
theory of creativity. Two years later he wanted to deal more extensively with
the complexity of the phenomenon, and published a comprehensive review of
the field with the most recent statements about his causal creativity model
(Eysenck, 1995). The book Genius: The Natural History of Creativity is a
masterpiece of clarity, and probably one of the best books he has ever
written—and that amounts to something. The Genius book exposes various
426 Further Eysenckian interests

positions on creativity and evaluates them in terms of theoretical importance

and degree of empirical support. However, rather than repeating all the
details, I will here first outline Eysenck's motives for writing the book, then
briefly discuss his creativity model in terms of selected details, and finally
evaluate his general approach to creativity.

3.2 In search of answers

Eysenck (1995) struggled for clear answers to questions like: "Can genius be
defined and measured?"; "Can creativity be defined and measured?"; "What
role does intelligence play in the development of either?"; "What is the
contribution of personality?"; "Is there any relation between genius and
"madness," and if so what is it?"; "Can we formulate a cognitive theory to
account for creativity, and describe the workings of the creative mind?"; "Can
we define and measure intuition, as one of the alleged characteristics of the
creative person?"; "What is the role of the unconscious, if any?" (p. 2). If a
genius-mad connection can be established, then " ... how does pathology
[causally] produce creativity? ... . Why ... how ... ?" (p. 123).
Eysenck proceeds in the best tradition of psychology to get the answers. He
settles for a cognitive theory about the workings of the creative mind, and looks
for the psychological factors promoting it. Eysenck prefers measures to
postulates, however imperfect, unlike so many psychologists. It worries him
that many psychologists search for "facts" without stating an a priori theory.
He therefore recommends that they first establish such a theory, however
fallible. He explicitly denounces the anecdotal-historical method, because it
exposes one to the risk of succumbing to errors of the unaided memory and
self-justifying introspections. He further sees a purely psychometric approach
as weak because it is more descriptive than explicative. What basically is
lacking in most creativity research is, in Eysenck's words: "... a proper
reference to the storehouse of knowledge accumulated by experimental
psychologists" (Eysenck, 1995, p. 6). His countermove is, accordingly, to draw
on psychophysiological, genetic, and psychopathological research on the brain,
creativity, and genius.

3.3 Eysenck's model for creativity

Eysenck's model for creativity is illustrated in Figure 20.1.
Briefly, most variables in the model are influenced by genes. The hippo-
campus is an important physiological formation for creativity. Dopamine
enhances, and serotonin reduces, trait creativity by directly affecting cognitive
inhibition. Dopamine thus reduces latent inhibition and negative priming,
thereby widening the associative horizon. Remote elements can then be easily
combined in a creative fashion. However, too high dopamine concentrations
 Molecular creativity, genius and madness 427

Motivational
and Cognitive
Variables (IQ)

Dopamine Latent Schizophrenia

Inhibition

f Creativity Creativity
(Trait) Achievment

Serotonin Negative Manic & Socio-

Priming Depressive Cultural
Illness Variables

* Genetic Determinants.
Figure 20.1. Eysenck's causal model for creativity (see text for explanation). From Eysenck (1995)
with permission.

link to functional psychoses and low creativity. The impact of increasing

dopamine is effectuated along a continuous spectrum, ranging from no
creativity to creativity to manic-depressive illness over schizo-affective distur-
bances to full-blown schizophrenia at very high concentrations. In other words,
too little dopamine results in enhanced latent inhibition and negative priming,
a steep association gradient, a narrow horizon for combining remote elements,
low creativity, low P score, and altruism. Too much dopamine leads to a high P
score, lack of inhibition, and negative priming, a flat association gradient, a too
wide horizon for orderly combination of remote elements, low creativity, and
acute psychosis. Moderate to high dopamine levels result in suitable cognitive
inhibition, a moderately tilted association gradient, an optimal association
horizon, fairly high P, high creativity, or genius. However, the high dopamine
level increases the likelihood of seeing milder psychopathology. Serotonin has
the opposite effects.
The left side of Eysenck's model explains, in other words, the circumstances
under which individual trait creativity comes to life. But even though creativity
as trait is a necessary basis, it is not a sufficient basis for proper creative
428 Further Eysencldan interests

achievement. The right side of the model therefore completes the story. In
addition to trait creativity, above average achievement requires an IQ above
120, the presence of personality traits like ego-strength and persistence, and
also suitable sociocultural circumstances.
An essential aspect of Eysenck's model is that the factors forming extremely
high creativity act synergistically rather than additively. This explains why
genius is so rare. The idea is that, if just one of the necessary factors is missing,
the product will automatically be only a little creativity and certainly no genius.
This is precisely what happens for most of us, according to the model. We may,
for example, be exceptionally good at combining remote elements, but if we
lack, say, the stamina or guts needed for an almost obsessive follow-up phase,
we will never become a genius. The multiplicative formula translates into a J-
shaped distribution for creative achievement hi a given population. A huge
number of nonachievers sleep at the bottom, some high achievers ascend, but
only a few isolated geniuses throne at the top.
Eysenck hastens to admit that his theory has not yet passed the acid test—a
direct study of genius in accordance with the model. We must at present,
therefore, remain satisfied with much encouraging indirect confirmatory
evidence emanating from several studies, including some of his own. The next
step amounts, in Eysenck's own words, to " ... a lot more work... before we can
hope to articulate a unified theory making possible serious efforts at
falsification" (Eysenck, 1995, p. 284).
It is easy to see why Eysenck's half psychological-half brain physiological
approach to creativity is superior to most psychological attempts. First, he
explicitly takes the scientific route and differs hi this from the many intuitively
based explanations of genius in terms of divine insight, talents, gifts, deep
contemplation, or dreamlike revelations. Eysenck prefers testable hypotheses
to the reified concepts some use on a purely descriptive basis as causal
variables to explain changes in other reified variables. Instead of undisciplined
speculations about multiple dimensions or a myopic focus on abstract or
hypothetical variables, Eysenck wants to deal with the task in terms of
measurable parameters of brain physiology. Eysenck never remains satisfied
with questionnaire data (although he certainly masters the technique),
correlations, purely descriptive psychometry, or factor analysis for the sake
of factor analysis. He insists on the full exploitation of the power of well-
designed experiments.

3.4 PROBLEMS WITH EYSENCK'S MODEL FOR CREATIVITY

If Eysenck admits that the holy grail of genius has not yet been found, then
what is still missing? Obviously, one thing is to acknowledge that genes play an
important role for creativity, but which genes, which proteins do they code for,
 Molecular creativity, genius and madness 429

where do these proteins go in the body and brain, and what exactly do they do
in the target tissues that affects creativity? Of course, Eysenck is well aware of
the problem. This is why he directs our attention to the great research
potentials in molecular genetics and biology. Another problem is to pronounce
dopamine the key factor linking genius, high P score, and increased risk of
psychopathology. Dopamine (and serotonin) is probably both heavily
implicated in all three, but is it the key factor or a covariate? The recent
progress in brain sciences confirms that the brain houses an immensely
complex, dynamic, truly interactive molecular chemistry, characterized by
multiple reciprocally interactive links between genes, neurotransmitter path-
ways, and the environment. Too narrow a focus on only one or a few
conspicuous agents or links may miss the intricacies of dynamic positive or
negative feedback, or feedforward mechanisms. Moreover, much important
molecular information is embedded in the release pattern and temporal
variations in concentration, inactivation, and receptor sensitivity. This means
that simple measures of absolute concentrations may not reflect the full or
interesting part of the biological effect of a chemical agent. Eysenck obviously
knows this, and his choice to at least start somewhere in a very complex brain
and then see where it leads makes perfect sense. Dopamine (and serotonin)
might actually not be a bad point of departure. However, accepting for the
moment the hypothesis that high dopamine concentration holds the key to an
understanding of high creativity, a new problem knocks on the door. Will the
dopamine hypothesis explain the remarkable sex-related differences in
creativity? As we ascend the ladder leading to genius, we observe still fewer
females. Close to the top there are almost no females, in particular in the more
technically based areas of science—those drawing heavily on visuo-spatial or
mathematical skills. The hypothesis implies that females must have, relatively
speaking, much less dopamine (or more serotonin) than males, and therefore
score lower P and creativity. They could also have much more dopamine,
according to the model, but this would mean a much higher P and many more
psychotic females than males. Is there such a marked sex difference in
dopamine, and does it explain the sex-related difference in creativity? A fourth
problem is to evaluate the proper (causal?) role of P in creativity (see chapter
6). Eysenck insists that his model is dimensional rather than categorial, and
that there is a continuum from common to creative to genius to mad. Along
that continuum runs an increasingly higher P score. To be sure, P reflects a
dispositional variable closely related to, but neither identical with creativity nor
with psychosis. Now, to which extent are creative individuals, and at least some
of their relatives, increasingly characterizable by traits like: criminal, impulsive,
hostile, aggressive, psychopathic, or schizoid (or, periodically, unipolar
depressive, affective disordered, or schizo-affective when not in a creative
phase, or downright schizophrenic and not ever creative), as opposed to
noncreatives showing conformist, conventional, emphatic, socialized, or truly
430 Further Eysenckian interests

altruistic traits. Eysenck hastens to regret that we simply have too little good
data to be sure, but it certainly is quite easy to point to families with
exceptionally creative historical or contemporary individuals, with an increased
incidence of psychoticism-like personality traits or psychosis (see chapters 6
and 19). Creativity relates in this way to schizophrenia (Eysenck, 1983, 1987,
1995; Heston 1966; Karlsson 1968,1970; McNeil 1971), and both Hammer and
Zubin (1968), and Jarvik and Chadwick (1973) suggested that there may be a
common genetic basis for great potential and for psychopathological deviation
(also see Rosenhan & Seligman 1989). Claridge (1990, see also chapter 17)
came to the conclusion that creativity may be more closely associated with
affective disorders than with schizophrenia, and Jamison (1989) found that
38% of all eminent British authors and artists needed treatment at least once
in their life for affective disorder. But then again, there are also examples of
healthy families with unaffected highly creative achievers (unless excessive
working style is defined as psychopathology). It is hard to know the exact
figures, because most reviews of the psychopathology of geniuses are prepared
from a particular perspective. The point is therefore not that P has little or
nothing to do with creativity, but rather that the connection may not be simple.
The psychometric peculiarities of the P scale may also muddy the water (see
section 6.6 in this chapter, and also chapter 17).
The perhaps most serious problem I see with the psychological approach to
creativity and genius relates to the notion of a (multiplicative) relationship
among causal factors. Like Eysenck, I see several good reasons for a
multiplicative rather than an additive relationship among the causal agents
producing genius, but it is the psychological approach that begs a crucial
question: What multiplies with what? Frankly, it makes no sense in precise
experimentally operationalizable terms to multiply a favorable brain dopamine
concentration with suitable cognitive inhibition and descriptive phenotypic
ego-strength, introversion, and dominance to get the product—high creativity.
To multiply fundamentally different chemical (molecular), infered
psychological (guessed) and descriptive psychometric (phenotypically
observed) factors amounts to committing Rylian category errors!
Obviously, this brief and highly selective excursion pays no justice to
Eysenck's general and overwhelming contribution to the area of creativity. The
narrow mission was to discuss basic elements in his model for creativity and to
see where it raises problems. The more general purpose was to show that his
research points to new and basically unmapped directions in an otherwise
stagnating research on creativity.
I will in the following reanalyze a series of instructive studies of exceptionally
creative scientists by Roe (1904-1991: see Wrenn, Simpson, Gorayska, & Mey,
1991), and then present a natural science model for individual development of
 Molecular creativity, genius and madness 431

ordinary, creative, genial, and mad states to account for the observations. I see
such a solution as just a radical extension of Eysenck's view on the science of
human nature.

4. ROE'S STUDIES OF EMINENT SCIENTISTS

Anne Roe was a versatile researcher and clinician, with a professional horizon
spanning from clinical work on schizophrenia to mapping personality
parameters of importance for various occupations to penetrating research
into exceptional creativity and achievement. Despite an obvious psychoanalytic
slant, she never forgot to look for data, and I for one have profited greatly from
the insight derived from her 1952-1953 series of studies of 64 scientists (1951a,
b, 1952a, b, 1953, 1970).
To see what is so special about the most eminent scientists in the U.S.A.,
Roe examined 22 physicists, 20 biologists, and 22 social scientists (psycho-
logists and anthropologists). They were all selected by a panel of experts to be
the best in their respective fields. The impressive list of honorable
memberships and prices awarded to them suggests that the judges made a
narrow selection.
Roe found large individual differences among these exceptionally creative
people and warns that a typical eminent scientist does not exist. She also
observed some interesting common features. Most were firstborn sons of
professional men. Almost all of them worked hard, devotedly, seven days a
week, almost to the brink of displaying an obsession. Many of them admitted
openly that their work is their life.
Clear group differences arose when the sample was categorized according to
discipline. Eminent biologists and physicists contrasted social scientists in most
respects (see below). Biologists and physicists tend to be shy and over-
intellectualizing. Many were sickly as a child, lonely, "different", and aloof
from their classmates. They were only moderately interested in girls, began
dating no sooner than college, and married on average at age 27, which is
rather late for national standards. Most of them continue to live in stable
marriages, behave unusually independently, and have few recreations. The few
recreations they have were typically those of a loner, such as fishing, sailing, or
solitary walks. They do not care much about family relations, and show little
guilt feelings about parental relations. They tend to avoid social affairs, parties,
political activities, and religion. Biologists rely strongly on nonemotional and
nonaggressive approaches to problems, and physicists show a good deal of free
anxiety. Roe found many exceptions, to be sure, but not enough to spoil the
general picture of an eminent natural scientist.
432 Further Eysenckian interests

The description of the behavioral scientists contrasted that of the physicists

and biologists in almost every conceivable way. Behavioral scientists tended to
be highly gregarious, and to be socially active at an early age. Often they were
acknowledged leaders already in school, where they practiced intense and
extensive early dating. They were deeply concerned with human relations,
showed many dependent attitudes, much rebelliousness, and considerable
helplessness. They tended to be quite openly aggressive, and to experience a
high divorce rate (41%).
Roe further noted that very few of these highly gifted scientists came from
the South of the U.S.A., none were Catholics, five came from Jewish homes,
and the rest were raised in Protestant homes. However, irrespective of
background very few scientists had any serious interest in religious matters.
Table 20.1 (from Nyborg, 1991) summarizes, in modified form, Roe's
observations of the overall pattern of representation of abilities and personality
in the different academic disciplines, and contrasts them with data for blue-
collar workers.
The table illustrates how abilities clearly distinguish natural from social
scientists. Roe, in fact, even found group differences within these
categorizations. To get that far, special tests to map exceptional verbal (V),
spatial (S), and mathematical (M) abilities had to be constructed by the
Educational Testing Service, as currently available standard tests were much
too easy for many of these eminent scientists. The physicists without question
scored highest on these demanding tests, but theoretical physicists performed
relatively better on verbal tests, and experimental physicists relatively better on
spatial and mathematical tests. Among the scientists, the biologists,
physiologists and botanists scored relatively higher on verbal, and geneticists
and biochemists relatively higher on nonverbal tests. Social scientists obtained
a significantly lower overall IQ score than physicists. However, even within this
group of scientists, social psychologists and anthropologists performed
relatively better on verbal tests, and experimental psychologists better on
spatial and mathematical tests. Some of the anthropologists were, in fact,
unable to understand the mathematical tasks, whereas the most difficult of
these items were too easy for some of the physicists. Here, perhaps, we have
identified an important factor in the differential developmental status and
sophistication of various scientific areas!
To summarize, eminent physicists and biologists tend to mature slowly, to
have a troubled youth, and to feel lonely, shy, and "different" as children.
Typically, they are not very interested in girls, many late, have few children,
and live stable solitary lives. They get very high IQ scores, but theoretical
physicists do better on verbal ability tests, and experimental physicists do better
on spatial tests and in mathematics. Social scientists mature faster, are more
popular, begin dating earlier, have more children, and are more likely to
Table 20.1. Body, brain, and behavioral development, and adult personality and ability patterns in exceptionally creative natural and
behavioral scientists (modified after Roe, 1952), and in blue-collar workers (from Nyborg, 1991b)

Natural sciences Social sciences Blue collar

Physicist Biologist Behavioral scientists occupations
Theoretical Experimental Physiologists/ Geneticists/ Social Experimental Unskilled
botanists biochemists psychologists/ psychologists workers
anthropologists
Body and brain
development
Early vs. late Late Late Late Late Late Late Early
i
».
Personality
Troubled youth + + + + - - + I
Shy and lonely
Early interest in
+
-
+
—
+
-
+
- + + + 1
girls
Stable family life + + + +
Emotionality — — — — + + + a,
Aggressiveness — — - — + + +
Ability ^
Verbal (V) V++ V+ V++ V+ V++ V+ V I
Spatial (S) S+ S++ S+ S++ S- S+ s-
Mathematical M+ M++ M+ M++ M- M+ M-
(M)
434 Further Eysenckian interests

become divorced. They get about as high verbal scores as the physicists, but
much lower spatial and mathematical scores. An exception is experimental
psychologists. They approach the intellectual pattern of the physicists.
Roe's work illustrates two important points. First, despite the many
similarities, extraordinarily high creativity or genius is not a unitary
phenomenon; it is to some extent domain-specific. Second, whether a given
individual will ever contribute anything creative in a particular domain
depends on his ability-personality constellation, everything else equal (e.g., a
suitable environment in the broad sense described later).

5. PHENOTYPIC SIMILARITIES AMONG CREATIVE AND HIGH IQ

INDIVIDUALS

Roe's description of the extraordinarily creative scientist dovetails nicely with

observations of high IQ people. This, actually, is not too surprising. Most of
Roe's subjects had IQs above 140, so we could expect to find at least some
analogies in the development of high IQ individuals and exceptionally creative
individuals. However, the argument to be developed here is not one of
complete identity. Creativity and IQ do not correlate above IQ 120, so we are
talking about some striking developmental similarities among high IQ and
creative individuals, but also a few decisive differences between creatives and
noncreatives to be addressed later.
What about sex-related development? High IQ individuals tend to mature
slowly, to become slightly taller, and to develop an androgynous body type,
relative to the average for their sex (e.g., K. B. Hoyenga & K. T. Hoyenga,
1979,1993; Nyborg, 1983,1994). The three groups of highly creative architects,
studied by MacKinnon (1962, 1964, 1970), showed an extremely high peak on
the Mf (femininity) scale of the MMPI. Studies of highly creative males by
Hassler, Birbaumer, and Nieschlag (1992), Kemp (1985), and others, confirm
that creative musicians tend to be characterized by psychological androgyny; so
do high IQ individuals (Maccoby & Jacklin, 1974).
Postwar (but perhaps not prewar) fertility, as measured by number of
offspring, is lower in high IQ individuals (Vining, 1982, 1984), but their life
expectancy is higher (Danmarks Statistik, 1985). There is a tendency for high
IQ boys to behave less physically aggressive, and for high IQ girls to behave
more physically aggressive than the average (Maccoby & Jacklin, 1974). Roe
(1952b) noted that exceptionally creative natural scientists tend to have few
children, social scientists more, but lower IQ.
What about sociability? Highly creative children in elementary schools tend
to feel estranged from their teachers and peers (Torrance, 1962), as do creative
adolescents (Getzels & Jackson, 1962) and high IQ children. Cattell and
Butcher (1968) found, like Roe, that adult research scientists tend to be
 Molecular creativity, genius and madness 435

skeptical, withdrawn, unsociable (McClelland 1962; Taylor & Barron 1963;

Terman & Oden 1959) critical, precise, apt to express socially rather
uncongenial and "undemocratic" attitudes (Van Zelst & Kerr, 1954)
associated with dominance (Rushton, Murray, & Paunonen, 1983; see also
chapter 19), to hold the belief that most other people are rather stupid, and to
show a surprising readiness to face endless difficulties and social discourage-
ment in order to have it their way. Barron (1965) finds that the original
individual rejects regulation by others, and has a strong need for personal
mastery, involving self-centeredness and self-realization. MacKinnon (1962,
1964, 1970) finds profound skepticism, rebelliousness, self-assertiveness, and
independency characteristic for highly creative architects, already manifested
clearly in school and onwards (Dudek & Hall, 1984).
Cattell and Butcher (1968) find, like Roe, that the typical research scientist,
and in particular the physical scientist, is introverted, stable, and withdrawn,
and characterized by a combination of higher than average ego-strength, high
anxiety, and excitability. Moreover, researchers are more self-sufficient, more
bohemian, and more radical than are successful administrators and teachers.
Cattell and Butcher further find greater susceptibility to nervous disorder
among artistic than among scientific geniuses, and that artists and literary men
are more bohemian and more emotionally sensitive, than are scientists, in
addition to having a higher ergic tension level and a general tendency for
greater instability and emotionality.
To summarize, Roe's exceptionally creative high-IQ scientists show a
number of similarities with creative high-IQ individuals observed in other
studies as well as with noncreative high-IQ individuals. Creative scientists of all
colors seem thus to have many important traits in common, but there is a
tendency for greater emotional instability and relatively speaking lower
intelligence in the social scientists.
These patterns of trait covariance leave us with a number of nagging
questions. Which proximal factors are responsible for the striking
developmental similarities between exceptionally creative scientists and
noncreative high-IQ people. Do these factors relate causally to measurable
brain parameters? Do we here find the mechanisms accounting for the
supposed lack of correlation between creativity and intelligence above IQ 120?
Phrased differently, what makes creative high-IQ people stand out from each
other and from noncreative high-IQ individuals? The answer to these
questions involves a proper solution to a crucial methodological question:
just how many analytic dimensions do we really need to account for the
developmental similarities and differences?
436 Further Eysenckian interests

6. COVARIANT TRAIT DEVELOPMENT

6.1 The General Trait Covariance (GTC) model
With respect to the question of proximal factors I have argued elsewhere that
gonadal hormones are ideally suited for coordinating sex-related development,
and formulated a General Trait Covariance (GTC) model with 12 principles to
account for hormone effects (Nyborg, 1979, 1983, 1984, 1994). Perhaps the
hormone principles could explain the above mentioned similarities between
creatives and other high-IQ individuals? Variations in hormone balances may
even account, at least in part, for domain-specific differences. In any case, the
original GTC model needs extension in order to formalize the harmonizing
and differentiating effects of genes, gonadal hormones, and environment on
creativity development (Nyborg, 1991a; b).
Briefly, the development of an originally sexually neutral fetus is guided by
three interdependent factors: genes, hormones, and experience. Hormone
production is thus determined by genes as well as by the environment.
Hormones exert organizational as well as activational effects on body and brain
tissues by modulating accessible genes in the genome, by affecting
neurotransmitter systems, and by changing cell membrane characteristics.
Gonadal hormones go everywhere in the body, but are biologically active only
in hormophilic tissues capable of inducing specific receptors for them. This
arrangement makes hormones uniquely suited to selectively co-ordinate and
pace body, brain, and behavioral development. The 12 principles account for
how the sexually neutral (except for Y chromosome material) fetus
metamorphoses into covariant male, female, or "something-in-between"
patterns of phenotypic traits.
Figure 20.2 presents a recent version of the original General Trait
Covariance (GTC) model for hormonally guided development (for more
details, see Nyborg, 1979, 1983, 1984, 1987b, 1988a, b, 1990a, 1991a, 1992a,
1994, 1995, 1997a, b, c).
The model works in the following way. Males and females are first divided
(somewhat arbitrarily) in different hormotypes in accordance with their
person-specific position on continuous androgen or estrogen dimensions. A
male with high plasma testosterone (t) is said to be hormotype A5, and a male
with low t is hormotype Al. A female with high plasma estrogen (£2: 17-/3-
estradiol) is hormotype E5, and a female with low £2 is hormotype El. The
advantage of hormotyping is, that the GTC model now generates rather
precise predictions about individual covariant body, brain, ability, and
personality development. It is actually a bit surprising to see how well the
predictions of covariant development fit available evidence (for reviews, see
Nyborg 1983; 1984; 1994).
 Molecular creativity, genius and madness 437

Males Females
Body Body

ANDROTYPE A5A4A3A2 A1 E1 E2 E3 E4 E5 ESTROTYPE

> Estradiol
Testosterone

Figure 20.2. The General Trait Covariance (GTC) model for development. The model generates
testable predictions about harmonized body, brain, intellectual, and personality development from
parental DNA, plasma testosterone/estradiol balance, and experiences. Optimum intellectual and
personality development is predicted by moderate and balanced hormone concentrations, but at
the cost of sexual differentiation. Maximum sexual differentiation is predicted by high and
contrasting testosterone and estradiol concentrations, respectively, at the cost of less than optimal
intellectual and personality development. However, even slight variations in sex-related hormone
exposure cause deviations from the expected modal pattern of male or female development,
because each trait has its own developmental trajectory, time table, and hormone receptor
sensitivity. The mechanism for this is probably, that hormones may transiently or permanently
"switch" parental genes on or off by modulating their transcription rate (after Nyborg, 1994).

6.2 Hormotypic similarity among creative high-IQ and noncreative high-

IQ individuals
Roe's descriptive studies, and those of others, raised the suspicion that creative
high-IQ and noncreative high-IQ individuals share common developmental
factors that make them deviate from the average person. Gonadal hormones
may be these proximal causal factors. Unfortunately, there are no direct
hormone studies of high-IQ people, so the hypothesis has to be evaluated in
terms of indirect evidence.
Confluent evidence suggests that extraordinary creative scientists and
noncreative high-IQ males both tend to have moderate to low plasma t
(Nyborg, 1991a, b). Let us tentatively assume that there are many hormotype
Al or A2 among them, and then follow the predictions of the model. These
males obviously are found in the upper left corner of the full spectrum of
hormotype—trait connections in Figure 20.2. The dashed line inside the curve,
guides us to their most likely modal development. Bodily, they tend to be
slightly taller than average (obviously, familial disposition plays a role here, but
then their parents probably were taller than average, too). They would also
have an above average fat/muscle ratio (due to catabolic effects of low t or
surplus E2, but again relative to family disposition). Als are expected to enter
puberty late and to show minimal early sexual interests, relative to the
population average. The low t Al male is predicted to be characterized by an
438 Further Eysencldan interests

androgynous sexual identity, with some "feminine" personality traits

interspersed among some not too dominant male traits, and by few social
interests (i.e., high introversion and occasional loneliness). He would
predictably father fewer children, live longer, and have a high threshold for
physical (but not necessarily verbal) aggression than the average. His
intellectual pattern is high V, high S, and high M skills, and he would thus
have a high Spearman's g. Even though these predictions for the low t, Al
high-IQ males were originally formulated without any thought on creativity,
they fit surprisingly well the picture Roe and others give of the exceptionally
creative natural scientists.

6.3 Hormones fine-tune differences in abilities and personality

Already the earliest version of the GTC model (Nyborg, 1979) predicted an
inverse relationship between S and V as a function of variation in gonadal
hormones. This was later confirmed in studies by Hampson (1986, 1990) and
by Hampson and Kimura (1988). This inverse relationship may provide a point
of departure for generating sets of hypotheses about the causal basis, not only
of within-group differences in abilities in extraordinarily intelligent people, but
also of group differences among creative natural and social scientists and
artists, and even of finer differences among subgroups of natural scientists. All
it takes is a few testable assumptions. Circa 50% of intelligence is co-
determined by genes; provided equal g, the V-gifted male has been exposed to
slightly more t than the S-M-gifted male; the gifted social scientists and artists
have been exposed to more t than the natural scientists; provided equal g, the
theoretically oriented high-V natural scientists have been exposed to slightly
more t than the experimentally gifted high S-M natural scientists.
Male social scientists may thus qualify as genetically gifted hormotype(s) A4
(or higher). In that case, the GTC model predicts the following trait
constellation. The slightly above average intelligent A4 male will be shorter
than the highly intelligent natural scientist, and have a lower fat/muscle ratio.
The more t would make him enter puberty earlier, make him more person
oriented, less shy, and more popular among peers than a prospective natural
scientist. His sexual identity would develop in a slightly more masculine
direction, associated with earlier awakening of sexual interests, though not to
the same extent as hi the A5 "macho" hormotype. He would marry more often
and have more children but, alas, would also divorce more often. He would die
earlier than his natural science colleague, due to an increased risk of high
blood pressure, circulatory diseases, heart attack, or prostate cancer. He would
be more prone to aggressive outfits, and react more emotionally and hostile
than the natural scientist, but less than the A5.
 Molecular creativity, genius and madness 439

There is some evidence suggesting that S and M abilities are more sensitive
to variations in gonadal hormones than is V (Gouchie & Kimura, 1990;
Nyborg, Nielsen, Naera, & Kastrup, 1992), even though V may be enhanced by
increases in t. We can accordingly expect that the slightly higher t (and slightly
lower E2) in hormotype A4 male, relative to A2s, would manifest itself in a
slight improvement in V (mostly in verbal fluency) and in a decrease in S and
M. This would go some way to explain the differences in patterned intelligence
between natural and social scientists (again, person-specific familial
dispositions obviously also have to be taken into account). The highly creative
artists may also be a moderately above average intelligent hormotype A4 or
higher.
Most blue-collar workers have definitely higher than average t (Dabbs, La
Rue, & Williams 1990; Nyborg, 1994), and would qualify as hormotype A4 or
A5 males. Their lower than average IQ can be explained in more ways. They
either carry a familial disposition for lower than average intelligence, or they
show a hormonal depression of a familial disposition for high intelligence.
Further deducible from Figure 20.2, they are expected to be shorter, more
athletic and muscular, and to mature earlier than average A3s and much
earlier than the more intelligent and the more creative A2 individuals
(obviously, again seen in the light of family dispositions). A5s will show very
early interests in girls, will display a more aggressive, impulsive, and person-
oriented style, and may socialize but condition less well than the average A3.
A4s and A5s will have more children, but will not live as long as A2s. Their
verbal fluency score may be higher than the P, S, and M scores, but then again
high t is particularly punitive for the expression of both P, S, and M. A5s
therefore tend to obtain lower than average scores on heavily g-loaded
intelligence tests.
There are not many exceptionally creative females. It is food for thought that
the few female artists with a well-deserved recognition as exceptionally
creative, show a tendency to write poetry and novels about people, or paint
flowers and other natural motives. The very few eminent females in natural
science are typically found in the biological disciplines. There are almost no
females qualifying as exceptionally creative mathematicians, musical maestros
or composers, chess masters, engineers, or architects. The GTC model suggests
that this is due to a slightly reduced P, S, and M score, relative to V, caused by
their relatively high E^t balance. The hormonal modulation of personality may
also be important for creativity. The lesser willingness of most females to
pursue a typical male obsessive pathway towards some remote goal for years
and years, and to pay what may seem to most women an unacceptably high
social and personal fee, may be related to their lower t. Let it be clear, that the
GTC model cannot provide final answers to the question of why so few females
pursue an exceptionally creative path. The model stresses an important point,
however, with respect to the importance of hormonal perturbations prenatally
440 Further Eysenckian interests

and at puberty. Even slight variations in the hormone balances in these two
periods may decisively alter the pattern of female (and male!) abilities and
personality in ways that make highly intelligent females (and males!) less likely
to join the creative elite or compete without compromises in the gruesome
power play for top positions. Note, however, that the model actually predicts
that some of the low-numbered estrotypes will make it all the way to the top.
El or E2 females, exposed to higher than average t (medically or physio-
logically, prenatally or at puberty, or briefly at menopause), may be more
inclined to begin or pursue a professional career path than average females
(e.g., Purifoy & Koopmans, 1979).
More generally, the model suggests that high levels of homotypic hormones
(t for males and £2 for females) are incompatible with scientific, artistic, or
occupational achievement. The high levels seem to depress the particular
abilities and personality traits called for in creative achievement in these areas.
Moreover, they elevate social inclination and caring attitudes in females, and
overt physical aggression in males. None of these traits are particularly
valuable assets in creative achievement. In very high (or low) doses hormones
may, in fact, not only disturb sensitive hormophile brain functions but can even
accomplish massive systematic neural cell death (Nyborg 1991b; 1992; 1994;
1997a).

6.4 Hormones fine-tune differences in brain structure and function

It has not been possible to identify with a sufficient degree of certainty the
specific brain structures subserving exceptionally high IQ and creativity (see
chapters 11, 12, and 14). Large areas of the brain are active during problem
solving, but there are large individual, regional, age- and sex-related
differences in task-related metabolic rates.
Hormones are important for the brain. In addition to coordinating body and
brain maturation, they affect neural plasticity and regulate the electric activity
of the brain. Let us consider the "simple" hypothesis that neural plasticity is
the missing link between low-numbered hormotypes and higher than average
creativity (Nyborg, 1991a; b).
The neural plasticity hypothesis for creativity has three components: a
temporal, a structural, and a functional. The temporal aspect refers to the
notion that an exceptionally intelligent adult brain is a brain that retains much
of its childhood plasticity long after puberty (for a critical discussion of
neoteny, see Nyborg, 1994, chapter 13). Neural plasticity thus becomes a
necessary, though not a sufficient, condition for above average intelligence.
The structural aspect refers to the notion that plastic brain tissues are more
likely than fully mature and solidly established tissues to reconfigure as a
function of use. The functional aspect relates (perhaps) to Eysenck's notion of
a wide association horizon.
 Molecular creativity, genius and madness 441

Hormones causally affect neural development, plasticity, and functionality,

so that only moderate prenatal and pubertal surges are compatible with high
adult intelligence and creativity.
Nottebohm (1981, 1989) has provided interesting animal support for the
hypothesis that gonadal hormones regulate the neural plasticity and associated
"creativity" in birds. Briefly, only males sing in many bird species, and then
almost exclusively during the mating season. They are then silent again until
next season. Nottebohm demonstrated that the neural song system of a bird is
highly sensitive to variations in plasma t. As t secretion increases gradually up
to the mating season, several neural song nuclei begin to form in the bird brain
and grow in volume, and probably also in number of neurons, and in richness
of synaptic connections. This gradual process at first allows the bird to draw
upon a primitive song repertoire with few vocals. With further increases in t,
the bird enters a period of "plastic song." When t is at its zenith it finally
reaches the full song stage with highly reliable song performance. This
coincides in time with the bird being sexually fully mature and in need of
effective means for attracting the attention of female birds for reproductive
purposes. The interesting part of this story is, that the "plastic song" stage is
the time, when the bird is most capable of learning new song variations. With t
at its maximum, the bird cannot but sing in a stereotypic way. Nottebohm
explains the plastic or creative song phase with optimum neural plasticity at
moderate levels of t. As t declines, and the mating season is over, there is a
reduction in the number of synaptic connections in the song nuclei, and the
bird's song repertoire becomes partly wiped out by "forgetting." A moderate
increase in t marks the overture to the next season, and allows new flexible
synaptic connections to be established, which enables the bird to create new
constellations of vocals and multicolored creative song.
Can Nottebohm's bird hormone model for neural plasticity serve as a human
model? What moderate t concentrations mean to bird-song nuclei and
potentials for "creative" song variations could roughly match what moderate
concentrations of gonadal hormones mean to human brain plasticity,
intelligence, and creativity. The hypothesis would provide a much-needed
temporal-structural-functional perspective on the Eysenckian notion of wide
association horizons in creatives. Most likely not only t per se, but also the
aromatization of t to £2 is involved, but an account of this complicated
technical story cannot be given here (see Nyborg, 1994, chapter 8)
Covariant body and brain maturation follows a person-specific develop-
mental timetable. Most children show natural ease in new learning, and often
combine old and new elements in "unexpected" ways throughout childhood.
Alas, this flexibility, and in particular the ability to combine elements in
unusual ways, often dissipates, in many to a dramatic extent, as children reach
puberty. Like for canaries, "childhood creativity" is supplanted by more robust
but sex-stereotypic adult performance, and the more so the earlier and larger
442 Further Eysenckian interests

the hormone surge. There are, to the best of my knowledge, no published

studies specifically addressing the question of covariant human brain plasticity,
intelligence, and creativity development during childhood. Presently, we have
to remain satisfied with bird and rat evidence, showing that high levels of
hormones negatively affect adult neural plasticity. Pavlides, Westlind-
Danielsson, Nyborg, & McEwen (1991) thus demonstrated, that neonatal
hyperthyroidism simultaneously disrupts hippocampal long-term potentiation
(LTP) and adult spatial learning. We took this to mean that the hormonally
conditioned reduction in neural plasticity disrupts the capacity to figure out
and remember where to find food in the eight-arm radial maze. Also relevant is
Shapiro's (1968) observation, that thyroid hormone treatment of neonatal rats
speeds up body and brain development and learning capacity before puberty
but, unfortunately, the early maturation and initial hormonal boost of learning
capacity has to be paid for with lower neural plasticity and inhibited learning
after puberty. Given that rat, bird, and human hormones and receptors are
chemically identical we might perhaps be justified in assuming that this is a
good animal working model for human development, the main difference
being, of course, that the hormones modulate partly identical/partly species-
different genes through similar mechanisms, in various animals.
In addition to the narrow hormone-brain plasticity hypothesis for creativity,
hormones could also account for broader covariant trait development by
monitoring the tempo of body and brain maturation, and in this way harmonize
body development with neural plasticity, intelligence, personality, and
creativity. High levels of pubertal hormones speed up body maturation,
reduce neural plasticity, hamper the expression of familial intelligence, and
reduce creativity. The hypothesis would account for the observed relationship
between early-late maturation and different patterns of intelligence and
creativity in Roe's samples of exceptional scientists in terms of variations in
gonadal hormones and their effects on neural plasticity. We know that high
hormone concentrations at puberty cause early closure of the growth zones in
the long bones, leading to low final body height. They also seem to reduce
neural plasticity. It is this covariance hypothesis that allows for quite specific
predictions from hormotype over brain mechanisms to which kind of children
will most likely suffer selective reduction of childhood creativity at puberty.
This application of the GTC model is discussed in more detail elsewhere
(Nyborg 1991a; 1994). It is a sad fact, however, that there are several
definitions of neural plasticity, each referring to complex and partly unexplored
brain conditions. Without a much better understanding of the details of the
hormone-brain connection we will not be able to test the hypothesis.
 Molecular creativity, genius and madness 443

6.5 What makes creative high-IQ individuals stand out from just high-IQ
individuals?
Section 6.2 demonstrated that high-IQ individuals share many developmental
characteristics with exceptionally creative high-IQ people that make both
groups differ from the average person. What makes creative high-IQ people
stand out from noncreative high-IQ individuals?
The tentative psychometric or factor-analytic answer is, that particular
differences in covariant combinations of intellectual and personality traits
make for the distinction, and that this also explains domain-specific differences
among scientific geniuses. This description is totally vacuous, however, unless
we can transcend the phenotypic surface description, and identify details in the
differences in various supporting molecular brain devices that lead up to
differences in psychometric intelligence, personality, and creative achievement.
To facilitate this explorative process, we better first recapitulate which
psychometric trait combinations fit which domains.
Unusual stamina, that is, an almost obsessive devotion to work despite
adversities is a must for any genius, according to Roe (1952b) and many others.
Above average intelligence is, too. Her study further indicates that eminence in
various disciplines requires different combinations of body and brain
development, abilities, and personality. Theoretical physics, physiology, and
botany may call for a combination of late maturation, introversion, and a V/S-
M balance. Experimental physics, genetics, and biochemistry calls for a
combination of late maturation, introversion, and an S-M/V balance.
Experimental psychology calls for the combination of moderately late
maturation, moderate introversion and IQ, and a S-M/V balance. Social
psychology and anthropology call for slightly above average intelligence,
slightly later than average maturation, moderately high extraversion, and a V/
S-M balance. Nonprofessional areas do not require high IQ, and neither early
maturation nor extraversion is a hindrance.
These phenotypic patterns equip us with a preliminary answer to the
question of what makes exceptionally creative individuals stand out from each
other, from noncreative high-IQ individuals, and from the average person. The
answer depends, in other words, on which area we are talking about, and at
least five different ability-personality combinations are needed to explain
Roe's between-group differences in Table 20.1. An interesting implication of
this is that—as the genius can neither select his own intelligence nor
personality nor his person-specific combination of abilities—we are forced to
conclude that the genius does not choose his scientific discipline at his own
discretion. It is rather the other way round: The domain "selects" him in
accordance with an evolutionary-like process involving selective pressures. If
his strong sides mismatch a particular domain, he and the domain waste time
and energy.
444 Further Eysenckian interests

The preliminary nature of this hypothesis of domain-specific selection for

genius is obvious. It might, nevertheless, facilitate the search for appropriate
causal models for creativity, genius, and madness. But what is an appropriate
causal model? Most likely, it is a model that accounts in strictly causal terms
for the complex processes and interactions behind the phenotypically
observable covariant patterns.

6.6 How many analytic dimensions are needed?

This raises a fundamental question: Is it possible—within the framework of a
single analytically and causally coherent model—to examine how particular
combinations of genes, hormones, neural plasticity, neurotransmitters, intel-
ligence, personality, and environmental circumstances play together to
produce domain-specific behavior like extraordinary intelligence and creativ-
ity? The solution to this problem presumes that it is possible to identify a
unitary causal level at which genes, hormones, and brain plasticity transform
intelligence and personality into creativity and genius, in the presence of an
ever-important environment. Neither traditional psychological analyses nor
contemporary gene-environment interaction analysis will allow us to go that
far (Nyborg, 1987a, 1989, 1990b, 1994, 1997a; Nyborg & B0ggild, 1989;
Wahlsten, 1990). Eysenck has, as usual, seen the problem already, and
recommends that we go beyond the empty lexical definitions, surface factor
analyses, and descriptive psychometry, and begin to look in the direction of
brain physiology and chemistry. In the following section I will take his advice to
its logical extreme, and settle for just one analytic dimension with three analytic
windows.

7. THE NONLINEAR, DYNAMIC, MULTIFACTOR, MULTIPLICATIVE,

MULTIDIMENSIONAL MOLECULAR (ND4M) MODEL FOR EXISTENCE
7.1 Introduction
The descriptive GTC model (Figure 20.2) was primarily designed to help
formalizing covariantly developing body and brain parameters in order to
become able to predict ability and personality patterns as a function of the
trinity of DNA-body chemistry-environment interaction (Nyborg, 1983, 1994,
1997a).
To progress, the GTC model was next extended to account in proximal
causal terms for the nonlinear, dynamic, multifactor, multiplicative, multi-
dimensional molecular interactions leading up to the existence of people and
all other existing configurations, and to their disappearance. The result of this
megalomane endeavor is a new ND4M model. A specialized version of the
model is presented in Figure 20.3, where it is adapted to focus narrowly on
human development, behavior, and society exclusively in terms of molecular
 Molecular creativity, genius and madness 445

Figure 20.3. A nonlinear, dynamic, multifactor, multiplicative, multidimensional molecular

(ND4M) model for common, creative, genial, or pathologic development. Creativity/sensitivity/
susceptibility is a combined function of particular gene constellations, moderate plasma sex
hormones, low sexual differentiation and high adult neural plasticity. Schizothyme development is a
function of low hormone and incomplete subcortical development, whereas cyclothyme
development is a function of high hormone and cyclic instability among molecular parameters.
The model mimics multidimensional mass molecular space-time-phase (jc, y, z + time + phase)
changes over long phylo- and shorter ontogenetic periods.

interactions. It is further restricted to illustrate how intelligence, creativity,

genius, psychopathology, and society can possibly be subjected to a single all
bottom-level molecular analysis. In that, the ND4M model differs radically
from the usual multilevel psychological approach as well as from the linear,
additive, and statistically based multivariate nature-nurture model for average
differences around a population mean.
The name of the model is indeed ackward, but each of the adjectives refers
to vitally important interrelated aspects of development. The following sections
first briefly describe what the adjectives refer to, and then discuss the
interaction with a emphasis on creativity, genius, and psychopathology.

7.2 A molecular account of human nature and society

It has been argued, to the horror of some psychologists and philosophers, that
(1) man is basically a molecular constellation, evolving, developing, acting, and
disappearing again like all other molecular constellations in a basically
molecular earthly world embedded in a larger molecular universe, that (2) all
446 Further Eysenckian interests

this can sensibly (although, of course not exhaustibly) be analyzed in terms of

molecules, and that (3) this is presently the only scientifically acceptable way to
try and account for covariant human (and societal) development. The research
program behind this view is presented in details, and related to a notion of our
exclusively molecular evolutionary past in Nyborg (1994).
Briefly, molecular DNA instructions are transcribed during ontogeny into
protein molecules that either amass intrasystemically into body and brain
tissues or affect their functionality when first formed. Environmental impacts
are more or less systematic changes in extrasystemic physico-chemical
parameters having an effect on intrasystemic molecular parameters through
the digestive and perceptual systems. One particularly important class of
extrasystemic impacts, social interaction (love, etc.), is defined as intersystemic
physico-chemical (molecular) interaction among systems made slightly
dissimilar by, among other agents, hormones. This exclusively molecular view
of man, society, and the universe is named physicology. The neologism refers
to a research program designed primarily to entangle molecular causes,
interactions, and effects in very complex systems, be that of organic or
inorganic origin. The presence or absence of carbon atoms is really not
essential to the analysis. The molecular level of analysis is chosen entirely for
practical reasons. Molecules have sizes and effects that place them
conveniently in between the remote small-scale elementary particle level and
the large-scale level of the cell or organs. Molecules are sufficiently close to the
human scale to be of practical value in the causal study of development and
function, whereas elementary particle physics would entangle the analysis in
the small-scale peculiarities of quantum mechanics. The cell structure level
would leave out of view many intricate fluid processes within the cell. Effects of
hormones can, for example, be analyzed entirely in terms of molecular
concentration, affinity, and time-space-phase coordinates if one wishes. So
can nerve cell membrane characteristics. Membranes are conglomerates of
molecules "frozen" temporarily in space in accordance with their stereotaxic
characteristics and environmental circumstances.
The left z-axis in the ND4M model in Figure 20.3 identifies the focus for a
given causal analysis. The first practical step in the analysis of systemic
molecular causes, interactions, and effects is to open an analytic window to
either the intra-, inter-, or extrasystemic aspects of molecular interactions.
Ideally, all windows should be opened at once, but presently we do not even
have proper tools for keeping track of everything going on in one.
Where the psychological analysis inevitably involves a tangled hierarchical
web of surface, top-down, and bottom-up analyses, in the futile attempt to
connect incompatible material and abstract spheres, the molecular account of
human nature involves a nonhierarchical all-bottom approach to the
examination of covariant molecular mass-actions separated more or less
clearly in space and time and defined by phase (see also chapter 25).
 Molecular creativity, genius and madness 447

7.3 Multidimensionality
The front #-axis in the ND4M model in Figure 20.3 provides an overview of
some of the metric state or trait parameters, that are often related to creativity
in the literature on genius. Three aspects of the model deserve mentioning in
that connection.
As already said, it is a very general model covering all aspects of becoming,
being, and breaking apart again (i.e., developing, living, and dying in the
animal and plant cases). Had the model not been adapted here to focus on
creativity, the metric indicators at the front *-axis would have had different
names, and would have referred to other molecular causal processes.
Second, the metric indicators are pure surface names with only descriptive
value. A common sin in the psychometrics of intelligence and personality is to
see such indicators as genuine causes, but this is a dangerous method of
explaining away what goes on intrasystemically. As mentioned previously, it
makes no sense in causal terms to say that genes interact additively or
multiplicatively with hormones, intelligence, and personality traits like ego-
strength, introversion, or social factors to produce intelligence or creativity.
Those who say so are multiplying apples with pears, and what do they get?
Third, the model circumvents the fatal category error problem inherent in all
hierarchical psychological, cognitive, or rational analyses. Each of the
descriptors in the ND4M model refers to a more or less well-defined
molecular mass-action process. These mass actions may share important serial
or parallel processual community, but the analysis still amounts to just one-
level molecular interaction in a truly cause-effect sense.
It would, for obvious reasons, be wrong to say that the physicological
program behind the model refers, in fact, to a traditional behavioral program.
Many of the decisive molecular interactions do not show up immediately in
phenotypic behavior, and some only much later in life. Physicology is therefore
rather a program for the study of the molecular dimensions behind behavior,
even though behavior is obviously an expression of interacting molecules
moving collectively in space-time coordinates. In other words, each descriptive
indicator tentatively lined up along the front *-axis of the ND4M model refers
to important events in a particular molecular dimension. The left y-axis
indicates the most likely space-time-phase coordinates for the interaction of
these molecular events.

7.4 Multiplicativity
The ND4M model is based on the notion of multiplicativity, quite like
Eysenck's model. The state or trait descriptors at the *-axis were selected
basically because they typify creative individuals or geniuses, and because none
of them is likely to be missing in the description. Creative achievement (right z-
axis) inevitably suffers if an individual has optimum scores for all state or trait
448 Further Eysenckian interests

descriptors at the Jt-axis apart from one. High Spearman g would, for example,
be wasted for creative achievement if ego-strength was missing (of course, this
obscure psychoanalytic term is taken here to reflect some kind of long-term
molecular consistency, perhaps related to t\). Neither can physico-chemical
factors like a suitable prenatal environment or nutrition be missing from the
formula for a genius. Even learning, memory, and social interaction can be
defined in molecular terms (Nyborg, 1994) and may then enter the molecular
formula for a genius.
Eysenck (1995, p. 49ff) assumes that intellectual and creative achievement is
best described by a J-shaped distribution (like so many other psychological and
socio-economical phenomena: Allport, 1934; Burt, 1943; Nyborg, 1991b;
Walberg, Strykowski, Ronai, & Hung, 1984). Price (1962; 1963) found,
however, that the distribution is better described in terms of an S-shape. This
makes sense, as nothing grows perfectly. I will follow Price's advice, and
inscribe the asymptotic multiplicative cause-effect on creativity in the ND4M
model in terms of an S-formed distribution. The total creative brain potential
is, accordingly, seen as an exponential product of a limited number of key
factors, each representing evolutionarily optimized molecular mass actions.
Some of the factors represent fairly stable aggregations of molecules forming
sensitive nerve cell membranes or other structures; other factors reflect mobile
neurotransmitters, peptides, or hormone-receptor complexes, and still others
represent robust DNA structures.
One implication of the notion of an S-shaped creativity achievement
distribution is that adding more and more optimally adjusted key factors to the
multiplicative formula means little to the expression of genius, as long as the
number of original key factors equals or surpasses an absolute lower number.
Further fine-tuning of factors in the existing genius may broaden his domain
specificity, however.
Another important characteristic of the model is that concomitant optimum
tuning of all factors is seen as a rare and probably rather unstable situation.
The loss or severe mistuning of just one key factor may spell a dramatic
deterioration in the performance of a genius. On the other hand, the extremely
rare occasion of a "divine stroke of genius" in an otherwise unremarkable
person may reflect a sudden lucky optimum tuning of all factors. Life-span
analyses may illustrate temporary shifts. Eminent physicists typically realize
their most creative potentials while young (sometimes even before the age of
25, even though Nobel prizes are typically given to old men). Age-related loss
of creativity may be due to, say, the inevitable (and highly regrettable!)
decrease in / production with age (e.g., Ellis & Nyborg, 1992). Of course, many
other age-related factors are involved, too. Perhaps the decreasing / levels
relate to loss of persistence (ego-strength or willpower, if you must!). Such
temporal shifts in molecular tuning could explain the sad fact that even the
most extraordinary creativity lessens a bit with time. On the other hand, a
 Molecular creativity, genius and madness 449

genius first flourishing at middle-age might actually reflect that he had too
much t to unfold full genius in young adulthood, where he overshot the
optimum. Elsewhere I hypothesized that most young adults loose their
"childhood creativity," because the considerable pubertal surge in sex
hormones reduces neural plasticity while enhancing sexual differentiation of
the body and brain (Nyborg, 1991a).
In summary, the absence of one or more key factors marks the difference
between genius and not genius. Fine-tuning of one or a few (ability or
personality?) key factor(s) explains the domain specificity of extraordinarily
creative individuals. More dramatic mistuning of one or more factors threaten
creative achievement in general. Although the model is basically multiplicative,
the tuning of single key factors affects the weight by which they enter the
formula for creativity and genius.
Which weight should be attributed to social factors in the fomula? Actually,
little if any, as there is no hard experimental evidence proving that unspecific
rearing or social engineering affects extraordinary creativity. The closest we
come to documentation for a socialization effect is Zuckerman's (1977)
observation that Nobel laureates tend to seek the working company of those
who already got the prize. However, this could either mean that clever domain-
specific people prefer the company of likes, or that creative role modeling
really works wonders. We simply do not know. The ND4M model nevertheless
remains fully open to any experimentally documented environmental effect,
though it has no space for loosely defined and poorly documented social
parameters like prevailing norms, cultural stereotypes, or passive role
modeling without a physico-chemical address. In contrast, factors such as the
chemical interaction between a pregnant woman and her fetus, nutrition, and
the modulation of neurotransmitters by stress, or learning from "significant
others" can and should be measured and entered into the multiplicative
formula for molecular interactions among factors (Nyborg, 1994). Eysenck, as
usual, strikes the truth when he says that much hard work remains to be done.

7.5 Creativity, psychoticism, and psychosis

Eysenck found an apparent paradox in the ^-creativity-psychoticism connection
in the GTC model (but he might not find it again in the ND4M model).
Eysenck's own creativity model associates high creativity with moderately high
dopamine and high P (and androgyny). The GTC model seems to associate
high IQ (and creativity) with low t, ample neural plasticity and low P, and low
creativity with high t, little neural plasticity and high P [females differ from
males in this respect: here, high creativity associates to high t (or low E2),
ample neural plasticity, and high P—see Nyborg, 1994—but this need not
concern us here].
450 Further Eysenckian interests

To explain the paradox, Eysenck wondered (1995, p. 276) whether genius is

the exception where "high testosterone levels and 'cognitive androgyny' may be
negatively correlated in the general population, but is closely associated in a
small sub-group of creatives." Perhaps so. The human brain actually
aromatizes some t to £2, there are large individual differences in aromatase
activity, and the mechanism might bear on Eysenck's suggestion. Unfor-
tunately, we know next to nothing about the possible effect of conversion on
neural growth and plasticity, brain function, creativity, and P. Another
possibility is that the brains of geniuses show reduced sensitivity of hormones.
However, many observations definitely speak against the high f-androgyny-
genius hypothesis. For example, high t relates negatively to both IQ and
introversion, and significantly so; high t further associates positively to early
somatic maturation, intense sexual and social interests, physical aggression,
and a stereotypic sexual identity. The covariant body-intelligence-personality
pattern of an A4 or 5 stands in contrast to the common description of a genius.
There is a way to solve the apparent paradox, however. Let us for a moment
contemplate the hypothesis that high P refers to different disorders in high and
low t males with high IQ. This hypothesis has two important implications. It
threatens Eysenck's notion of a smooth dimensional causal continuity between
normals, affectively disordered, and schizophrenics. It might partly resolve the
previously mentioned problem of the psychometric irregularity of the P scale.
To see how, we have to simplify complex matters considerably.
Using the nonlinear molecular dynamics of the ND4M model, I propose the
following three developmental hypotheses:
1. Extremely low prenatal and pubertal t disposes for slow body development
characterized by incomplete sexual differentiation, and a vulnerable neural
development with abnormal migration and/or incomplete (primarily
subcortical?) dendritic aboreation. The result is incomplete (sub)cortical
development, a tendency for enlarged ventricles already at birth,
abnormally high neural plasticity, and/or nervous instability and sensitivity
(and perhaps high dopamine), with resulting confused thinking and
disturbed perception, as seen in schizophrenia. Low t is, according to this
hypothesis, made partly responsible for the tendency of male schizo-
phrenics to mature late and show a neotenic and somewhat demasculi-
nized sexual development, with little interest in girls, a tendency for
attaining a linear body build (as noted by Kretchmer), extreme
introversion and reduced contact with reality, and high P score.
2. The less extreme cases of low t males would still be characterized by slow
neural and somatic development, but now associated with optimum
migration and dendritic aboreation, unusual synaptic connectivity, and
optimum neural plasticity, sensitivity and a wide association horizon (read:
molecular covariation). The lucky combination of suitable parental DNA-
 Molecular creativity, genius and madness 451

moderate hormone exposure could dispose for the effective development

of a large brain with above average intelligence (at least above IQ 120),
and result in the creative hormotype Al or A2 natural scientist with higher
M-S than V scores seen in Table 20.1. This creative low-? individual might
nevertheless earn a high P score. Extreme sensitivity, due to an exceptional
overall brain state with optimally tuned parameters, may already in
childhood result in eccentric behavior matching that of a prospective
noncreative postpubertal schizophrenic patient. Negative reactions from
significant others may further stress such a child enough to raise his P
score. Obviously, this speculation calls for chemical rather than psycho-
analytic testing.
3. A male with DNA favorable for high IQ, but now combined with the much
higher prenatal and/or pubertal t exposure of an A4 would also earn a high
P score, although now for quite different reasons. The relatively high t
would dispose for brain development characterized by general neural
overexitation, some neural plasticity, sensitivity, and creativity, but also for
early body and brain maturation, extraversion, and a social inclination.
This could be the genesis of the social scientist or artist in Table 20.1, with
a higher V then M-S balance. Perhaps ^-related neural overexitation
represents, in extreme cases, an unstable brain condition alternating
between mania and depression. This would explain why a surprisingly large
number of eminent high V authors suffer from affective disorders and
score high P. The fact that acute affective psychotic states relate to periods
with low creativity comes as no surprise.
The hypotheses suggest, in other words, that creative social scientists, artists,
and authors may score as high on the P scale as the creative natural scientist,
but for entirely different endocrine and neural reasons. This interpretation
obviously does not fit Eysenck's dimensional idea of an underlying continuum
from normality over affective to schizophrenic conditions. It rather suggests
that P, creativity, and psychopathology covary as a function of hormones and,
of course, genes and enviromental factors. One thing is sure: There are
presently too few good data to take a firm stance in the matter.

7.6 The molecular dynamics of the ND4M model

The ND4M model is build on molecular dynamics, and this is the basis upon
which the version of the model accounts for the development, continuity, and
the eventual disappearance of creativity and genius. This means that creativity
and genius are seen as states rather than traits, even if the states prevail for a
long time, given stable molecular circumstances. Extraordinary creativity, or
something like it, can be observed in some eccentric children before puberty, as
their brains have the considerable neural plasticity and capability needed to
452 Further Eysenckian interests

combine remote processes in unexpected and sometimes productive ways.

However, large pubertal surges of sex hormones may at first speed up the
tempo of maturation, but then put an end to long bone as well as brain growth
potentials. The higher the surges, the sooner the termination of body and brain
growth, and final neural plasticity. The creative child will remain creative after
puberty only if he or she is exposed to low to moderate hormone con-
centrations prenatally and at puberty, or if the brain was primed prenatally by
hormones to low sensitivity to adult hormone concentrations. Whether
childhood "creativity" gets a further boost with brain growth at puberty or
will be inhibited depends, in other words, on the right gene-hormone
concentration combination, but also on hormone-binding globulins, receptor
sensitivity, and a favorable environment. Only neotenic children with moderate
amounts of sex hormone will retain their childhood "creativity." Some data
speak in favor of this hypothesis (Hassler, Birbaumer, & Nieschlag, 1992).
Other data also suggest that molecular brain processes subserving IQ are
sensitive to hormone molecules: Spearman g is definitely negatively correlated
with t in males (Nyborg, 1994), as are visuo-spatial abilities in high £2 females
(Hampson, 1986, 1990; Hampson & Kimura, 1988; Nyborg, 1979, 1983).
The dynamic aspects of the ND4M model extend far beyond puberty. The
model predicts, for example, that a woman will show slightly enhanced
creativity shortly after menopause, relative to her creativity during the
reproductive period. The material basis for this prediction is straightforward.
When ovulation stops, the pituitary reacts to the drop in plasma E2 with
increased gonadotropin release. This stimulates the adrenals to secrete more
substances with androgenic effects for a couple of years. Many postrepro-
ductive females have, in fact, relatively speaking quite high androgen and low
E2 status (lower than many men of comparable age!), and this ought to show
up in a short-lived increase in physical energy, nonverbal IQ and, according to
the model, creativity. Hormones further affect brain processes of relevance for
personality parameters. As t drops with age (Ellis & Nyborg, 1992), we can
expect less neural plasticity and disturbed molecular brain processing, lower
ego-strength, dominance, psychoticism, and perseverance. Loss of one or more
of these factors means loss of the state of genius, and a dramatic reduction in
creativity for the rest of us.
As said before, the ND4M model and the physicological research program
presume that the last variable to the right on the Jt-axis, the environment, is as
much a purely physical-chemical parameter as is the transcription of DNA
material. Environmental molecular parameters like prenatal fetal exposure to
maternal hormones or viral infections, birth complications, nutrition, stress of
all kinds, systematic changes in molecular brain parameters caused by
perception or learning, and intersystemic (social!) interaction must all find
their proper place and weight in the formula for creativity and genius.
However, social conditions in general, and systematic creativity training in
 Molecular creativity, genius and madness 453

particular, must be rigorously defined in operational terms before they can be

allowed to enter the formula for genius. This is evidently not the case now, and
there is even some evidence to suggest that opposing social conditions may
stimulate some geniuses to work even harder.
This brief discussion is not meant to cover the many dynamic possibilities for
enhancing or inhibiting molecular brain processes of relevance for phenotypic
creativity. It suggests, however, that it might be worth our while considering the
brain as a complex molecular system at the brink of instability (Nyborg, 1997a).
In fact, this may be the only scientifically acceptable way to approach creativity,
genius, madness, and organic existence in general.

7.7 Nonlinearity
Molecular actions and reactions typically unfold in nonlinear interactions in
most biological systems. A slight increase in the concentration of a given
chemical species gives a linearly graded response, but further increases often
result in nonlinear responses. Very high concentrations may turn the effect
into its opposite or into something quite different.
The state of genius thus presumes rare DNA combinations predisposing for
optimum flexible brain development and functioning. Genes for moderate
hormone secretion, and a favorable environment (prenatal or otherwise—not
exposing the fetus to unusually high or low levels of natural gonadal hormones,
stress hormones, or artificial hormones), is also a must, as is moderately low
pubertal hormone secretion. Abnormally low hormone concentrations
negatively affect neurogenesis; moderate hormone levels relate to optimum
neural plasticity; high hormone levels to overactivation of neural tissues.
Studies, summarized in Nyborg (1984; 1990a; Nyborg et al., 1992) even suggest
that the general karyotype (XX or XY) is less important phenotypically than is
hormone exposure with respect to covariant body, brain, intelligence, and
personality development, even though in most cases karyotype and hormotype
go together. E2 may, for example, feminize the brain in weak concentrations,
masculinize it in larger doses, and have neurotoxic effects in high doses. The
duration of exposure is also important. For example, short-term increase in
stress hormones may have beneficial effects, but long-term surges in stress
hormones may cause systematic cell death in sensitive brain tissues.
The ND4M model connects the nonlinear molecular effects with phenotypic
behavior in fairly simple ways. The different layers in Figure 20.3 refer to
differences in the tuning of the various molecular systems. A person's level is
determined by covariant interactions along the left y-axis among the many
nonlinear molecular subsystems lined up along the front jc-axis as metric state
or trait descriptors. Intermediate effects can for convenience be expressed in
broad terms like genotype, hormotype, neurotype, and phenotype along the
right y-axis. Starting with DNA transcription, the first factor to the left on the
454 Further Eysencldan interests

x-axis, protein production by specific genes may be much too high or low to
benefit brain structures subserving the personality of the genius, or it may
disturb proper functioning of critical brain areas underlying, say, intelligence.
In either case the DNA transcription factor would be missing in the
multiplicative formula, the end product of the molecular formula would be
zero and there would be no genius. For cases, where the optimum molecular
levels are either over- or undershot, the level of achievement would approach
the second or third levels. The state of genius (first level) is actualized only in
the extremely rare case where all molecular processes play pretty close to the
optimum at each of the inverted U-curves. Proper DNA transcription of
proteins will then relate to optimum hormone balance, ample brain
development with rich neural plasticity, and other suitable modes of
molecular functionings subserving the genius. Most of us either under- or
overshoot the top of one or many of the curves, and our creativity therefore
hovers somewhere in the area between the second and third levels.
Multiplicativity, nonlinearity, affinity, and space-time coordinates are the
tools by which the model accounts for the molecular dynamics of creativity, and
genius is the rare case where all parameters are optimally tuned. Domain-
specific differences among genius, like those observed by Roe (Table 20.1)
arise if one or a few of the parameters, such as hormone balances, varies
slightly around the optimal fine-tuning, furthering in some cases V, in other
cases M or S abilities, and at the same time covariantly fine-tuning body and
personality parameters.
The third layer in the model represents an inhomogenous majority. Some
may suffer from familial transmission of genes not favorable for the
development of an intelligent brain; others may be the victim of unhappy
emergenetic recombinations; still others may secrete too few or too many sex
hormone molecules to fit creative development. These, plus many other
conditions such as accidents and illness explain low third layer achievement.
Hormones have been attributed much weight in the previous sections. It is
worth remembering, however, that they too are only intermediary buffers in
the complex interplay between DNA, the brain and the environment—quite
like dopamine and serotonin.

8. THE FUTURE

Creativity research has so far been dominated by three major approaches: the
anecdotal-historical, the psychometric, and the psychological-sociological. The
harvest of studies conforming to these traditions is not impressive. Serious
problems remain, the explanatory power is low, and some specialists in the
area of creativity now talk about signs of degenerating research. Then Eysenck
entered the scene and suggested that experimental and physiological tools
 Molecular creativity, genius and madness 455

supplemented the psychological approach, and research began to move again.

Eysenck's creativity model generates testable brain hypotheses, instead of
trying to excuse failing predictions, and this is a significant improvement.
My only quarrel with Eysenck's approach is that he is not going far enough
in the right direction. His hybrid cognitive-physiological approach still keeps
too close to that of a classical psychologist to comfort, even if, over time, it
looks more and more like a bottom-up than a surface or top-down approach
characteristic of much contemporary psychology. More precisely, what really
worries me is that largely undefined (and perhaps ultimately undefinable)
terms like mind, cognitive inhibition, and ego-strength still play an important
role in a causal network, so that abstract mental and solid material variables
feature side by side in a multiplicative model for creativity, genius, and
madness.
With the recent progress in the brain sciences, the time may be right to skip
the uncensored use of hypothetical psychological constructs and intervening
variables. We can now begin to take our first faltering steps along perhaps the
only proper scientific avenue to the study of creativity, genius, and madness,
namely, in terms of all-bottom causal analyses. This becomes possible only
when DNA, the body and brain, intelligence, personality, creative achievement,
and the environment are all defined in terms of molecular mass interaction,
We need no a priori theory to accomplish this. All we have to do is to map or
guide where molecules go, and then see what they do when they meet—for
example, when leading up to or away from the states of genius or madness.
This may actually be the only acceptable definition of becoming, being, and
going away again. It is very fortunate that we do not have to map the fate of
each molecules for this, because that would have made the task entirely
impossible. Identifying differences in stereotaxic affinity, concentrations,
uptake, and biological action, plus real-time picturing or mathematical or
molecular modeling of mass-concentration effects in time-space coordinates
may suffice.
The physicological research program and the GTC and ND4M models are
based on this view. Even though Eysenck thinks favorably of it, he finds that I
go too far (Eysenck, 1996). He may be perfectly right, but so help me my
molecules, I see no other way around. Any solution that tries to combine
abstract psychic or cognitive with material factors is bound to sink into
intractable body-mind problems and trap the researcher into committing
inexcusable category errors. Physicology steers free of both kinds of problems,
by first resolving that the body-mind problem is a philosophical
pseudoproblem, and then by resorting to a unitary all-bottom analysis
(Nyborg, 1994). There is no reason to deny, however, that the physicological
research program faces a serious problem. This problem has less to do with
theory than with methodology. Even the most sophisticated contemporary
natural science methods cannot deal properly with complex nonlinear
456 Further Eysenckian interests

(molecular) dynamics. Where the futile anachronistic reference to abstract

psychic, mental and cognitive entities, philosophical body-mind divisions, and
the unlimited generation of hypothetical causal variables have impeded the
behavioral science, including creativity research, for centuries, also physicology
will stall as long as we have access only to rather primitive tools for
representing minute variations over time in the DNA-biochemistry-
environment formula for people, the environment, society, and existence in
general (see chapter 25). Progress in (creativity) research now depends
critically on our ability to examine and control the nonlinear dynamics of
molecular processes. We must become able to simulate data-dense real-time
molecular mass-action processes by massive parallel/serial computing,
simplifying graphics or, preferably, for real, or we might not be able to
clearly see the nuts and bolts in the processes leading up to states we call social,
intelligent, creative, genial, or mad. Eysenck's Genius book shows, that he is
one of the most glorious fighters here, allowing his inner "Catherine wheel" to
spin and spark once again, and ready to move as fast as ever in new directions.

REFERENCES

Albert, R. S. (1992). Genius and eminence (2nd ed.). Oxford: Pergamon.

Allport, F. H. (1934). The J-curve hypothesis of conforming behaviour. Journal of Social
Psychology, 5, 141-83.
Amabile T. M. (1983). The social psychology of creativity. New York: Springer.
Barron, F. (1965). The psychology of creativity. In: New directions in psychology, II. New
York: Holt, Rinehart & Winston, 1-134.
Burt, C. L. (1943). Ability and income. British Journal of Educational Psychology, 13,
126-140.
Canter, S. (1973). Some aspects of cognitive function in twins. In: G. S. Claridge, S.
Canter, & W. I. Hume (Eds.), Personality differences and biological variation: A study
of twins. Oxford: Pergamon.
Cattell, J. McK. (1903). A statistical study of eminent men. Popular Science Monthly, 62,
359-77.
Cattell, R. B. & Butcher, H. J. (1968). The prediction of achievement and creativity.
Indianapolis: Bobbs-Merrill.
Cattell, R. B. & Drevdahl, J. E. (1955). A comparison of the personality profile (16 P.
F.) of eminent researchers with that of eminent teachers and administrators, and of
the general populations. British Journal of Psychology, 46, 248-61.
Claridge, G. (1990). Great wits and madness. In: R. S. Albert (Ed.), Genius and
eminence (2nd ed.) (pp. 329-350). New York: Pergamon.
Cox, C. (1926). The earfy mental traits of 300 geniuses. Stanford: Stanford University
Press.
Dabbs, J. M., La Rue, D. de, & Williams, P. M. (1990). Salivary testosterone and
occupational choice: Actors, ministers, and other men. Journal of Personality and
Social Psychology, 59, 1261-1265.
 Molecular creativity, genius and madness 457

Danmarks Statistik (1985). Kvinder & Maend (Women & Men). Copenhagen:
Danmarks Statistik.
Dudek, S. Z. & Hall, W. (1984). Some test correlates of high-level creativity in
architects. Journal of Personality Assessment, 48, 351-359.
Ellis, L. & Nyborg, H. (1992). Racial/ethnic variations in male testosterone levels: A
probable contributor to group differences in health. Steroids, 57, 72-75.
Eysenck, H. J. (1983). The roots of creativity: Cognitive ability or personality trait?
Roeper Review, 5, 10-12.
Eysenck, H. J. (1987). Book review of D. N. Jackson & J. P. Rushton's (Eds.), Scientific
excellence: Origins and assessment. Personality and Individual Differences, 9(2), 447-
448.
Eysenck, H. J. (1989). Die Bewerting der Kreativitat mit Hilfe des Psychotizismus-
Werte. In: R. Lindner (Ed.), Einfallsreiche Vemunft-Kreativ durch Wissen oder
Gefiihl? Zurich: Edition Interfrom.
Eysenck, H. J. (1993a). Creativity and personality: Word association, origence and
psychoticism. Creativity Research Journal, 7, 209-216.
Eysenck, H. J. (1993b). Creativity and personality: Suggestion for a theory. Psycho-
logical Inquiry, 4, 147-178.
Eysenck, H. J. (1995). Genius: The natural history of creativity. Cambridge: Cambridge
University Press.
Eysenck. H. J. (1996). Special Review of "Hormones, sex, and society: The science of
physicology," Personality and Individual Differences,. 21(4), 631-632.
Eysenck, H. J. & Eysenck, S. B. G. (1975). Manual of the Eysenck Personality Question-
naire. London: Hodder & Stoughton. San Diego, CA: EDITS.
Galton, F. (1869/1978). Hereditary genius. New York: Julian Friedmann.
Getzels, J. W. & Jackson, P. W. (1962). Creativity and intelligence. New York: Wiley.
Ghiselin, B. (1952). The creative process. New York: Mentor.
Glover, J. A., Ronning, R. R., & Reynolds, C. R (Eds.) (1989). Handbook of creativity.
New York: Plenum Press.
Goertzel, M., Goertzel, V., & Goertzel, T. (1978). Three hundred eminent personalities.
San Francisco, CA: Jossey-Bass.
Goertzel, V. & Goertzel, M. G. (1962). Cradles of eminence. London: Constable.
Gouchie, C. & Kimura, D. (1990). The relationship between t levels and cognitive ability
patterns. Psychoneuroendocrinology, 16(4), 1-30.
Guilford, J. P. (1950). Creativity. American Psychologists (Vol. 5), 444-454.
Hammer, M. & Zubin, J. (1968). Evolution, culture and psychopathology. Journal of
General Psychology, 78, 154-175.
Hampson, E. (1986). Variations in perceptual and motor performance related to phase
of the menstrual cycle. Canadian Psychology, 27, 268.
Hampson, E. (1990). Estrogen-related variations in human spatial and articulatory-
motor skills. Psychoneuroendocrinology, 15, 97-111.
Hampson, E. & Kimura, D. (1988). Reciprocal effects of hormonal fluctuations on
human motor and perceptuo-spatial skills. Behavioral Neuroscience, 102, 456-459.
Hassler, M., Birbaumer, N., & Nieschlag, E. (1992). Creative musical behavior and sex
hormones: Musical talent and spatial ability in the two sexes. Psychoneuroendocrinol-
ogy, 97(1), 55-70.
458 Further Eysenckian interests

Heston, L. L. (1966). Psychiatric disorders in foster home-reared children of

schizophrenic mothers. British Journal of Psychiatry, 112, 819-825.
Hoyenga, K. B. & Hoyenga, K. T. (1979). The question of sex differences: Psychological,
cultural, and biological issues. Boston, MA: Little, Brown and Company.
Hoyenga, K. B. & Hoyenga, K. T. (1993). Gender-related differences: Origins and
outcomes. Boston, MA: Allyn and Bacon.
Jackson, D. N. & Rushton, J. P. (1987). Scientific excellence: Origins and assessment.
Newbury Park, CA: Sage.
Jamison, K. R. (1989). Mood disorders and patterns of creativity in British writers and
artists. In: R. S. Albert (Ed.), Genius and eminence (2nd ed.) (pp. 351-356). New
York: Pergamon.
Jarvik, L. F. & Chadwick, S. B. (1973). Schizophrenia and survival. In M. Hammer, K.
Salzinger, & S. Sutton (Eds.), Psychopathology. New York: Wiley.
Karlsson, J. L. (1968). Generalogic studies of schizophrenia. In D. Rosenthal & S. S.
Kety (Eds.), The transmission of schizophrenia. Oxford: Pergamon.
Karlsson, J. L. (1970). Genetic association of giftedness and creativity with
schizophrenia. Hereditary, 66, 177-182.
Kasparson, C. J. (1978). Psychology of the scientists: XXXVII, Scientific creativity: A
relationship with information channels. Psychological Reports, 42, 6491-6494.
Kemp, E. (1985). Psychological androgyny in musicians. Bulletin of the Council for
Research in Music Education, 85, 102-108.
Lombroso, C. (1901). The man of genius (6th Ed.). New York: Charles Scribner.
Lykken, D. T. (1982). Research with twins: The concept of emergenesis. Psychophysiol-
ogy, 19, 361-73.
Lykken, D. T., McGue, M., Tellegen, A., & Bouchard, T. J. (1992). Emergenesis.
American Psychologists, 47, 1565-1577.
Maccoby, E. E. & Jacklin, C. N. (1974). The psychology of sex differences. Stanford, CA.:
Stanford University Press.
MacKinnon, D. W. (1961). The creative person. Berkeley, CA: University of California
Press.
MacKinnon, D. W. (1962). The nature and nurture of creative talent. American
Psychologist, 17, 484-495.
MacKinnon, D. W. (1964). The creativity of architects. In: C. R. Taylor (Ed.), Widening
horizons in creativity (pp. 259-378). New York: Wiley.
MacKinnon, D. W. (1970). The personality correlates of creativity: A study of American
architects. In: P. E. Vernon (Ed.), Creativity. London: Penguin Books.
Mansfield, P. S. & Busse, T. V. (1981). The psychology of creativity and discovery.
Chicago, IL: Nelson Hall.
McClelland, D. C. (1962). On the psychodynamics of creative physical scientists. In H.
E. Gruber et al. (Eds.), Contemporary approaches to creative thinking (pp. 141-174).
New York: Atherton.
McNeil, T. F. (1971). Prebirth and postbirth influence on the relationship between
creative ability and recorded mental illness. Journal of Personality, 39, 391-406.
Nichols, R. C. (1978). Twin studies of ability, personality, and interests. Homo, 29, 158-
173.
Nottebohm, F. (1981). A brain for all seasons: Cyclical anatomical changes in song
control nuclei of the canary brain. Science, 214, 1368-1370.
 Molecular creativity, genius and madness 459

Nottebohm, F. (1989). From bird song to neurogenesis. Scientific American, February,

56-61.
Nyborg, H. (1979). Sex chromosome abnormalities and cognitive performance V:
Female sex hormone and discontinuous cognitive development. Paper and hand-out
for the symposium on "Cognitive Studies" at the Fifth Biennial Meeting of the
International Society for the Study of Behavioural Development, Lund, Sweden, 25-
29 June.
Nyborg, H. (1983). Spatial ability in men and women: Review and new theory. Advances
in Human Research and Therapy, 5, 39-140.
Nyborg, H. (1984). Performance and intelligence in hormonally-different groups. In
G. D. Vries, J. D. Bruin, H. Uylings, & M. Corner (Eds.), Sex differences in the brain.
Progress in brain research (pp. 491-508). Amsterdam: Elsevier Biomedical Press.
Nyborg, H. (1987a). Individual differences or different individuals? That is the question.
The Behavioral and Brain Sciences, 10, 34-35.
Nyborg, H. (1987b). Covariant trait development across races and within individuals:
Differential K theory, genes, and hormones. Paper presented in the symposium on
"Biology-Genetics" at the Third Meeting of the International Society for the Study of
Individual Differences, Toronto, Canada, 18-22 June.
Nyborg, H. (1988a). Change at puberty in spatioperceptual strategy on the rod-and-
frame test. Perceptual and Motor Skills, 67, 129-130.
Nyborg, H. (1988b). Sex hormones and covariant body, brain and behavioural
development. Neuroendocrinology Letters (Abstracts), 10, 217.
Nyborg, H. (1989). The nature of nature-nurture interaction. Behavior Genetics, 20,
738-739.
Nyborg, H. (1990a). Sex hormones, brain development, and spatio-perceptual strategies
in Turner's syndrome. In D. Berch & B. Bender (Eds.), Sex chromosome
abnormalities and human behavior: Psychological studies. Boulder, CO: Westview
Press.
Nyborg, H. (1990b). Good, bad, and ugly questions about heredity. Behavioral and Brain
Sciences, 13, 142-143.
Nyborg, H. (1991a). A model for selective sex hormonal depression of creativity at
puberty. Paper presented at the XXIInd Congress of the International Society of
Psychoneuroendocrinology, Sienna, Italy, 17-20 June. Neuroendocrinology Letters
(Abstract), 13(3), 187.
Nyborg, H. (1991b). Development of exceptional scientific creativity. Paper presented at
the XXIInd Congress of the International Society of Psychoneuroendocrinology, Sienna,
Italy, 17-20 June.
Nyborg, H. (1992). Genes, hormones, maturation, and the ethiology of psychoses: An
application of the GTC-A/E model. Paper presented at the International Symposium
on "Rate of Maturation, Brain Development, and Behaviour," University of
Trondheim, Geilo, Norway, 29 March-3 April.
Nyborg, H. (1994). Sex, body, mind, and society: The physicological approach. New York:
Praeger.
Nyborg, H. (1995). Intelligence and personality is when genes, hormones, and
experience exchange molecules in body and brain. Paper presented at the Vllth
Meeting of The International Society for the Study of Individual Differences, Warsaw,
Poland, 15-19 July.
460 Further EysencJdan interests

Nyborg, H. (1997a). Nonlinear gonadal hormone modulation of the brain and behavior
along the life cycle. In: U. Halbreich (Ed.), Gonadal hormones, sex, and behavior
(chapter 9). New York: American Psychiatric Press (in press).
Nyborg, H. (1997b). Limits to equality? The Danish experience (in preparation).
Nyborg, H. (1997c). Personality, psychology, and the molecular wave: Covariation of
genes with hormones, experience, and traits. In J. Bermudez, B. de Raad, A. M.
Perez, A. Sanchez-Elvira, & G. L. van Heck (Eds.), Volume of personality psychology
in Europe. Tilburg, The Netherlands: Tilburg University Press (in press).
Nyborg, H. & B0ggild, C. (1989). Mating behavior—moves of mind or molecules? The
Behavioral and Brain Sciences, 12, 29-30.
Nyborg, H., Nielsen, J., Naera, R. & Kastrup, K. W. (1992). Sex hormone therapy
harmonizes body, brain, and ability development and restores visuo-spatial ability in
young girls with Turner's syndrome. Paper presented at the 39th Annual Meeting of
the American Academy of Child and Adolescent Psychiatry, Washington, DC, 20-25
October.
Ochse, R. (1990). Before the gates of excellence: The determinants of creative genius.
Cambridge: Cambridge University Press.
Pavlides, C., Westlind-Danielsson, A. I., Nyborg, H., & McEwen, B. S. (1991). Neonatal
hyperthyroidism disrupts hippocampal LTP and spatial learning. Experimental Brain
Research, 85, 559-564.
Price, D. (1963). Little science, big science. New York: Columbia University Press.
Price, D. J. (1962). The exponential curve of science. In B. Barber & W. Hirsch (Eds.),
The Psychologist, 47, 1565-77.
Purifoy, F. & Koopmans, L. (1979). Androstenedione, testosterone, and free
testosterone concentration in women of various occupations. Social Biology, 26,
179-188.
Radford, J. (1990). Child prodigies and exceptional early achievers. New York: Harvester.
Roe, A. (1951a). A psychological study of eminent biologists. Psychological Monographs:
General and Applied, 65, No. 331.
Roe, A. (1951b). A psychological study of physical scientists. Genetic Psychology, 43,
121-239.
Roe, A. (1952a). A psychologist examines 64 eminent scientists. Scientific American,
187, 21-25.
Roe, A. (1952b). The making of a scientist. New York: Dood, Mead.
Roe, A. (1953). A psychological study of eminent psychologists and anthropologists,
and a comparison with biological and physical scientists. Psychological Monographs:
General and Applied, 67, No. 352.
Roe, A. (1970). A psychologist examines 64 eminent scientists. In P. E. Vernon (Ed.),
Creativity (8th edn). Middlesex, U.K.: Penguin.
Rosenhan D. L. & Seligman, M. E. P. (1989). Abnormal Psychology (2nd edn). New
York: Norton.
Rushton, J. P., Murray, H. G., & Paunonen, S. V. (1983). Personality, research,
creativity, and teaching effectiveness. In R. S. Albert (Ed.), Genius and eminence
(2nd edn.) (pp. 281-301). New York: Pergamon Press.
Shapiro, S. (1968). Some physiological, biochemical, and behavioral consequences of
neonatal hormone administration: Cortisol and thyroxine. Comparative Endocrinology,
10, 214-228.
 Molecular creativity, genius and madness 461

Sternberg, R. J. (Ed.) (1985). The nature of creativity. Cambridge: Cambridge University

Press.
Taylor, C. W. & Barren, F. (Eds.) (1963). Scientific creativity: Its recognition and
development. New York: Wiley.
Terman, L. M. (1925). Genetic studies of genius: Vol. I. Mental and physical traits of a
thousand gifted children. Stanford, CA: Stanford University Press.
Terman, L. M. & Oden, M. H. (1947). Genetic studies of genius: Vol. IV. The gifted child
grows up. Stanford, CA: Stanford University Press.
Terman, L. M. & Oden, M. H. (1959). Genetic studies of genius. Vol. V. The gifted group
at mid-life. Stanford, CA: Stanford University Press.
Torrance, E. P. (1962). Guiding creative talent. Englewood Cliffs, NJ: Prentice-Hall.
Van Zelst, R. H. & Kerr, W. A. (1954). Personality self-assessment of scientific and
technical personnel. Journal of Applied Psychology, 38, 145-147.
Vernon, P. (1982). Creativity: Selected readings. New York: Penguin.
Vining, D. (1982). On the possibility of a reemergence of a dysgenic trend with respect
to intelligence in American fertility differentials. Intelligence, 6, 241-264.
Vining, D. (1984). Subfertility among the very intelligent: An examination of the
American Mensa. Personality and Individual Differences, 5(6), 725-733.
Wahlsten, D. (1990). Insensitivity of the analysis of variance to heredity-environment
interaction. Behavioral and Brain Sciences, 13, 109-161.
Walberg, H., Strykowski, B. F., Ronai, E., & Hung, S. S. (1984). Exceptional
performance. Review of Educational Research, 54, 87-112.
Wrenn, R. L., Simpson, E. L., Gorayska, B. E., & Mey, J. L. (1991). Anne Roe (1904-
1991). Obituary. American Psychologist, 5(1), 116-126.
Zuckerman, H. (1977). Scientific elite: Nobel laureates in the United States. New York:
The Free Press.
Chapter 21
Eysenck's personality theory and organizational
psychology

A. Furnham

1. INTRODUCTION

Few psychologists have demonstrated such a range of research interests as

Hans Eysenck. From a Ph.D. in psychological aesthetics he has researched and
theorized in areas as diverse as behavior genetics, personality, intelligence,
social attitudes, psychotherapy, behavior therapy, sexual and marital behavior,
smoking and health as well as astrology and parapsychology (S. Modgil & C.
Modgil, 1986). Yet he appears to have been relatively little interested in
occupational and organizational psychology. A glance at the contents page or
index of either his biography (Gibson, 1981) or his autobiography (Eysenck,
1990) yield almost no references to issues in organizational psychology like
motivation, selection, absenteeism, or work productivity.
Perhaps the first question to ask is whether it is indeed true that Eysenck has
neglected occupational and organizational psychology? The second question is,
if indeed it is true, why that should be so? It does appear that none of his many
books directly deals with organizational issues. But there is a scattering of book
chapters (Eysenck, 1953) and papers (Eysenck, 1967) that deal specifically with
organizational psychology.
His early interest seemed concerned mainly with selection. Thus in the
classic Uses and Abuses of Psychology (Eysenck, 1953) one of the four sections
was labeled "Vocational psychology" and had four essays entitled "From each
according to his ability;" "The use of tests in student selection;" "Assessment
of Men;" and "Work, productivity and motivation." Indeed one primary
interest in the application of his theory is still in the world of selection (Barrett,
Kline, Paltiel, & Eysenck, 1996). Hence the conclusion of a section called
"Occupational performance" in an excellent comprehensive text book:
In sum, it appears that preferences for different kinds of occupation and occupational
success are both determined to some extent by personality. The research to date
mostly suffers from the disadvantage that job characteristics are discussed in an ad
hoc fashion. A major dimension along which jobs can be ordered is the extent to
 Eysenck's personality theory and organizational psychology 463

which the behavior of an individual doing that job is constrained by external factors.
For example, a car worker on an assembly line has minimal control over his work
activities, whereas a university lecturer has great control. It seems likely that
personality will be a more consequential determinant of job satisfaction and success
when severe constraints exist. It may be no coincidence that two of the occupations
wherein personality has been found to be relevant (flying and driving) both involve
considerable constraints. In other words, the fit of a worker to his job is especially
important when the worker has little scope for tailoring the work environment to his
needs. (H. Eysenck & M. Eysenck, 1985, pp. 328-329)
Yet it does seem the case that in the remarkable cannon of Eysenckiana
there are very few studies looking at work psychology. Why is this the case? It
is certainly not due to the fact that he does not believe in applied research. As
was noted in Cohen (1977) who interviewed Eysenck:
Do you feel psychologists have a duty, then, to do work that is useful to society?
"No. I don't think that one should ever prescribe for other people what they should
do or shouldn't do. I do think that many psychologists are interested in pure science
and the best of luck to them. It's an important thing for people to do. I'm glad people
are doing it. I have a kind of... I'm not sure what... a sense of duty to repay society in
some degree for the leisure it gives one and the chance to pursue one's fancies and so
I feel that some of the work I do should be useful, at least. It would go a little way to
repay society, which was why I worked in behaviour therapy, though it didn't wildly
fascinate me, but it was useful. I don't think everyone else should feel that way. It's
just a personal feeling." (p. 123)
In fact, Eysenck has done a fairly large number of important occupational
and organizational studies. He has, however, only mentioned these "in
passing" such as in the foreword to Furnham (1992a).
On one occasion, I was asked by a large firm of motor-car manufacturers to test
applicants for the position of apprentices in the firm, positions which were widely
sought after because of the high wages and security enjoyed by workers in that
industry at the time. I devised such a battery, and it was used for quite a number of
years to the satisfaction of the company. However, when I suggested that the success
of the battery should be tested against the actual performance of the apprentices, the
company politely refused. They said that they were quite satisfied and didn't see any
need for such validation. I pointed out that surely some apprentices would do better
work than others, and that it would be worthwhile to correlate these differences in
performance against the selection tests. They said: "No, all the apprentices are doing
equally well", which obviously makes no sense at all. However, there was nothing to
be done; I still believe that our battery of tests was a good predictor, but company
policy made it impossible to prove that hypothesis! There is widespread dislike in
many companies of the need to assess performance, and when it is done, it is usually
done so poorly and subjectively that the results are practically worthless. Hence, it is
often difficult to prove the efficacy of the selection procedure, not because the
procedure itself is not objective and valid, but because the criterion is poor or absent.
464 Further Eysenckian interests

Another problem that often occurs is the unrealistic nature of the requirements
stated by a company. When I first engaged in selection procedures in the United
States, I asked the Chairman of the company about the kind of people he wanted me
to select. "Well, I certainly don't want 'yes' men; I want people who are independent,
creative and original in their work." When I devised a battery that would procure
such people, together with the type of personality that is associated with creativity
and novelty-producing temperament, he was highly displeased and finally returned to
a retinue of "yes" men! Bosses don't always know what they really want, and what
they have to say is more often motivated by stereotypes that have no basis in fact.
This problem is related to another one, one which besets the psychologist engaged
in selection. What a given company needs, and what it wants, may be two quite
different things. I remember being asked to look at the selection procedures used for
the Civil Service, and to comment about their adequacy. I had to say that they were
well designed to produce the same kind of person who was already running the Civil
Service, so that my report was very favourable. It was not part of my brief to say that
possibly the kind of person who was running the Foreign Office was in fact unlikely to
make the right decisions, and had certain generalized attitudes and beliefs which
made his decisions unlikely to meet the needs of modern society. The psychologist
selector usually works to a brief, and it is not part of his business to dispute that brief!
For many companies, that is what is most urgently needed.
Another problem for the selector is the fact that most bosses believe that they
know as much about psychology, or more, than the people they employ as
psychologists. One of my students was once asked to discover why the products of a
certain manufacturer concerned with the production of mints were relatively
unsuccessful, particularly when comparing his sales with those of Polo Mints
(Lifesavers to our American friends!). My student interviewed customers and sales
people and came back with a very simple answer—people got more for their money
buying Polo Mints than buying his products! He exploded with fury. "It's obvious
what they are responding to—it's the sexual symbolism of the Polo Mint with the hole
in the middle! He fired my student, and hired a psychoanalyst instead. He went into
liquidation a year later. There is a general tendency for leading business people to
accept psychological advice only when it agrees with their prejudices, whatever these
may be. (pp. xv-xviii)
In fact, Hans Eysenck with colleagues and students, has not only been
interested in organizational psychology questions but has done a significant
amount of research in the area. The reason, however, why so little is published
is that much of this research was commissioned by organizations who forbade
publication of the data lest they get into the hands of the opposition.
Furthermore, Eysenck never had the easy introduction to the business/
commercial world that he had to the clinical world (personal communication).
Hence, as Claridge (1986) points out, Eysenck's work is inextricably linked with
abnormal psychology.
Yet as many have noted, Eysenck's theory is particularly fecund and
generates many predictions which have direct relevance to the world of work.
For 40 years researchers have relied upon parts of Eysenck's personality theory
 Eysenck's personality theory and organizational psychology 465

to generate and test hypotheses. (Cook, 1985). Thus although Hans Eysenck
himself has not personally published much work in occupational and
organizational psychology journals, his students, admirers, and distracters
have. Furthermore many have found the personality theory extremely useful to
devise and test specific hypotheses.
They have been helped by Eysenck (1967a; 1971) who reviewed work
showing that extraversion is linked with vocational preferences and various
aspects of industrial performance. The diverse literature shows extraverts
display greater "social intelligence" than introverts, that is, the ability to relate
to other people, to take a personal interest in them and their problems, and to
anticipate their reactions. Extraverts tend to gravitate towards and perform
best in, jobs that involve dealing with other people (e.g., sales and personnel
work, nursing, and teaching). The ability of the introvert to resist boredom and
persist with a task for a long period of time is also valuable in other
occupational contexts. Introverts are more reliable and conscientious, they are
more punctual, absent less often, and stay longer at a job (having less need for
novelty). While on the job, the extraverts appear to waste more time talking to
their workmates, drinking coffee, and generally seeking diversion from the
routine.
This review will consider some of the applications of Eysenck's personality
theory to the world of work.

2. EARLY STUDIES USING THE MPI AND EPI

Thirty-five years ago Rim (1961), one of Eysenck's students looked at

personality determinants of job incentives. He found students scoring low on
extraversion and neuroticism ranked "opportunity to learn new skills" as more
important than high scorers, while high neuroticism scorers ranked "good
salary" as more important than low scorers.
Bendig (1963) used the SVIB with the MPI questionnaire and discovered
that introverts preferred scientific and theoretical jobs such as journalism,
architecture, and the teaching of mathematics, whereas extraverts expressed
more interest in occupations involving social contact (e.g., selling life insurance
and social work). Extraversion was consistently and negatively correlated with
preferences to become an architect, dentist, mathematician, physicist,
engineer, or chemist, while neuroticism was negatively correlated with
accountant, office manager, banker, sales manager, and teacher.
In a study more useful for its norms than theory, Eysenck (1967b) collected
EPI data on 1504 businessmen. His results are shown in Table 21.1.
466 Further Eysencldan interests

Table 21.1. Personality scores of businessmen in different areas of business

Neuroticism Extraversion Lie scale

Area of business N Mean SD Mean SD Mean SD
General management 165 7.04 4.03 11.13 3.58 2.80 2.57
Production 135 6.90 3.77 11.05 3.72 3.08 1.69
R&D 574 7.42 4.05 9.98 3.88 2.76 1.49
Finance 132 7.53 4.49 10.12 3.40 2.93 1.97
Sales 168 7.04 3.64 11.33 3.98 2.92 1.92
Personnel 88 7.11 4.04 11.34 4.36 2.95 1.72
Consultancy 218 7.32 3.93 10.09 3.93 2.91 1.66
More than one of above 24 7.70 5.16 11.91 3.26 2.66 1.49

Total 1504
Standardized data
Normal population 2000 9.06 4.78 12.07 4.37
Salesmen 37 8.38 4.72 13.63 3.76
Professional 23 7.95 5.11 11.40 4.91
Normal population 651 - - - 2.61 1.57

Source: Eysenck (1967b).

On the Extraversion scale, the business groups are relatively introverted, but
significantly different between themselves, with finance, R&D, and consultants
being the most introverted, and those belonging to more than one group being
the extroverted. Eysenck noted:
Successful businessmen are on the whole stable introverts; they are stable regardless
of what type of work they do within business, but their degree of extraversion may be
related to type of work. The data are probably reasonably reliable because relatively
few respondents failed to answer, and because scores on the lie scale did not indicate
any market tendency to "fake good". The results suggest that the E.P.I, may have
some modest role to play in furthering research into the personality patterns of
person engaged in business and industry (p. 250).
The notion that introverted workers are better able than extroverted ones to
handle routine work activities was investigated by Cooper and Payne (1967) hi
a study carried out hi the packaging department of a tobacco factory where the
work was repetitive and light. Job adjustment, as assessed by two supervisors,
was negatively related to extraversion, and those workers who left the job in the
12 months following testing were significantly more extroverted than those who
remained. Neuroticism was also implicated, being related to poor job
adjustment and to frequency of nonpermitted absence. They note:
Beginning with the withdrawal indices, we find that the only appreciable correlations
are with length of service and non-permitted absence. The more extroverted workers
in this study have shorter periods of service to their credit than the less extroverted
(more introverted); this finding may be taken as evidence that the more extroverted
individuals will withdraw permanently from work of a routine nature. Non-permitted
 Eysenck's personality theory and organizational psychology 467

absence offers further interesting support for the withdrawal assumption. The
correlation between extraversion and surgery attendance, although in the expected
direction, is probably too small to merit serious attention. Surprisingly, certified
sickness absence is almost completely unrelated to extraversion; it would be tempting
to account for this non-relationship on the basis of certified sickness requiring a visit
to a doctor and subsequent submission of a medical certificate to the employer, all of
which may not be considered worth the effort when there exists the alternative of
taking one or two days' uncertified absence (i.e., non-permitted absence) with
virtually no trouble at all. However, such an explanation is not in keeping with an
unpublished finding of Taylor that extraversion scores for 194 male oil refinery
workers correlated .22 with sickness absence, (p. 112)

The use of both conditioning and arousal theory is evident in Cooper and
Payne's thinking. Because extraverts condition poorly and introverts readily,
extraverts are less able to tolerate tasks of a routine nature since inhibition
accumulates and inhibits sustained task performance. Also because extraverts
are underaroused they seek arousal and do not function as well as introverts
with a minimal or moderate sensory variation input
Savage and Stewart (1972) also found that 100 female card-punch operators
in training showed negative correlations between extraversion and supervisor
ratings of output per month, although there was no relationship between this
personality variable and drop-outs from the program. Following this theme,
Hill (1975) compared the behavior of introverts and extraverts on a
monotonous task. He found, as predicted, that extraverts tend to build more
variety into their responses on a monotonous task compared to introverts.
Looking at the selection and job-person fit literature Wankowski (1973)
found that extroverted students tended to choose practical or people-oriented
courses, whereas introverted students preferred more theoretical subjects.
Introverts had greater examination success than extraverts in the physical
sciences. Low neuroticism scorers opted for practically biased courses, whereas
high neuroticism scorers preferred people-oriented courses. In terms of
examination success, low neuroticism was associated with success in the
applied sciences.
Wilson, Tunstall, and Eysenck (1972) used various ability and personality
tests (including the EPI) to predict three criteria among gas fitters:
examination results, supervisor ratings, college attendance. Neuroticism was
a much better predictor than extraversion and results showed high scores were
negatively associated with both exam results and absenteeism. This finding
appears to be consistent across all occupations.
Rim (1977) got several job applicants to complete the EPI and rated
statements according to how well they described their ideal job. Among the
male subjects, the neurotic extraverts had the most distinctive ratings, valuing
social contact, economic and social position, patterning of time, and power
468 Further Eysenckian interests

functions of work more than neurotic introverts, stable extraverts, or stable

introverts. There were only modest and uninterpretable effects of personality
on the description of the ideal job among female subjects.
Since neurotic individuals in general, and neurotic introverts in particular,
are especially susceptible to stress, it might be thought that such people would
prefer jobs that involve minimal stress. However, Rim (1977) did not find any
large differences in the ideal job as a function of either neuroticism or neurotic
introversion, while Bendig (1963) reported only that high neuroticism was
associated with a dislike of business-type occupations such as banking, office
management, and accountancy.
Organ (1975a, b) examined personality correlates of conditionability in
organizations as operationalized by students getting bonus points for perfor-
mance on random tests. Introverts did better than extraverts, who presumably
got diverted from the routine discipline of the daily preparation for classes
regardless of contingencies.
Extraverts are more likely than introverts to prefer occupations that involve
social contact. There is therefore a danger that introverted workers may
become overaroused if their jobs involve considerable extra organizational
contact and a relative absence of routine. Blunt (1978) argued that introverted
managers would thus tend to choose positions involving relatively routine
duties (finance, production, or technical managers), whereas extraverted
managers would be more likely to select jobs in sales, marketing, or transport.
The results were broadly as hypothesized, except that transport managers were
less extraverted and production managers more extraverted than predicted.
Kirton and Mulligan (1973) found attitudes towards organizational change
to be related to a combination of neuroticism and extraversion among 258
managers from eight companies with at least 1000 employees each. Subjects
scoring high on both neuroticism and extraversion, and subjects scoring low on
both scales (neurotic extraverts and stable introverts) had more positive
attitudes toward change in managerial practices in general, more positive
attitudes towards specific, innovative appraisal schemes or promotional
policies being introduced, and the lowest level of discontent with the
institution and with superiors.
But do personality scores predict productivity and success? Turnbull (1976)
found that among more than 100 college students involved in a summer of
book sales, neither EPI extraversion scores alone, nor in combination with
other personality scales predicted sales success. Sales success was determined
on the basis of total wholesale business and a sales index indicating amount of
business per call made. In the global studies presented earlier, the sales
vocations were only weakly related to extraversion. Turnbull noted a wide
range of scores on the extraversion-introversion dimension among the
individuals applying for the job, and no personality differences between
those who completed the summer of sales and those who dropped out. It was
 Eysenck's personality theory and organizational psychology 469

found, as predicted, that extraversion scores increased from the beginning of

the summer to the end of the summer as a result of sales experience, an
increase that was equal for more successful and less successful salesmen.
Kim (1980) using undergraduates on a simulated work task found, as
predicted, that introverts were less dissatisfied on a nonstimulating task than
extraverts who were more satisfied on a simulating task, although there was a
difference in their actual performance. It was also found that introverts and
extraverts differed in their perception of expectancy and motivating
characteristics of objective tasks.
Studies of personality correlates of mood have revealed interesting results.
Christie and Venables (1973) asked 80 volunteers, whose jobs ranged from
office clerks to heads of academic departments in various schools, to complete
a mood adjective checklist on Monday and Friday mornings and afternoons for
four successive weeks. They combined the scales of concentration, activation,
and deactivation to form an efficiency index, for which there was a significant
four-way interaction effect involving the day of the week, the time of day,
extraversion, and neuroticism. The authors described a pattern of high arousal
and low euphoria experienced by neurotic introverts on Monday morning to
Monday absenteeism, and a pattern of high arousal and high euphoria
experienced by stable extraverts on Friday afternoon to premature departures
from work at that time. Bishop and Jean Renaud (1976) related end-of-day
moods to amount of change in daily activities and personality. Choosing people
at random in a community, representing a number of different vocations, they
asked subjects to keep a diary in which entries were made each 15 min during
both a work day and a leisure day. Mood ratings were taken from the last hour
before bedtime. Again, there was a four-way interaction effect. Activity
variation was not related to mood on work days, but it was on leisure days. The
fact that activity variation was related to pleasantness of mood for stable
extraverts and neurotic introverts but related to unpleasantness of mood for
neurotic extraverts and stable introverts indicates how increased variation and
stimulation (and its opposite, monotony) has different value for different
individuals.
In a study of personality correlates of job preference and satisfaction, Stern,
Alexandra, Barrett, and Dambrot (1983) found that extraverts preferred jobs
with higher levels of cognitive task demands, pace of task demands, cognitive
closure, extrinsic rewards, and intrinsic rewards. Neuroticism, on the other
hand, was negatively related to each of these preferences, except for extrinsic
rewards. Extraverts were less satisfied with clerical work itself, supervision, and
co-workers than introverts. It should be pointed out that subjects were
nonmanagerial civil service clerical employees in a job that would suit stable
introverts more than extraverts.
470 Further Eysenckian interests

These "early studies" did find considerable evidence for the arousal and
conditionability processes that underlay Eysenck's theory of extraversion and
neuroticism. However, many were piecemeal and there is always the concern
that the results are not generalizable because of the highly specific work-
related variables measured.

3. PERSONALITY JOB FIT

The arousal and conditionability process that is part of Eysenck's theory means
that it is possible to test some obvious ideas about the suitability of particular
personality types to specific jobs. As an example, the work on pilots and police
officers will be discussed.
Fairly impressive findings were obtained among training pilots by G. Jessup
and H Jessup (1971) who tested would-be pilots with the EPI early in their
course and discovered that the subsequent failure rate varied considerably as a
function of personality. Specifically, 60% of the neurotic introverts failed,
against 37% of the neurotic extraverts, 32% of the stable extraverts, and only
14% of the stable introverts. Thus, high levels of neuroticism had a much
greater adverse effect on introverts than on extraverts. They note that they
expect the introverted cadet to learn better both in the aircraft and lecture
room than extraverts. Jessup and Jessup concluded:
The comparative failure of the specifically neurotic introvert may be tentatively
explained as follows. High arousal in the visceral system is associated with high N;
high cortical arousal with low E. Given that there is no optimal level of arousal for
learning to fly and that this is a particularly stressful experience, it seems likely that
the neurotic introvert will be aroused beyond optimum; the learning of the stable
introvert on the other hand profits from cortical arousal with suffering from
additional visceral arousal, (p. 120)
Similar findings were reported by Reinhardt (1970), who carried out a
battery of personality tests on a sample of the U.S. Navy's best pilots. Their
mean score on the neuroticism scale of the MPI was only 11, compared with a
mean of 20 among American college students. Okaue, Nakamura, and Niura
(1979) divided the extraversion and neuroticism scores of military pilots into
three categories (high, average, and low) on each dimension. Of the sample of
75 pilots, 38 fell into the stable extravert category, with the highest frequency in
any of the other eight categories being only 8. In research with military pilots in
the U.K., Bartram and Dale (1982) found a tendency for successful pilots to be
more stable and more extraverted than those who failed flying training. They
had data on over 600 pilots from the Army Air Corps (AAC) and the Royal Air
Force. The consistent findings that neuroticism is negatively related to flying
success makes intuitive sense. Flying can obviously be stressful, with a single
mistake proving fatal. In such circumstances, pilots who are especially
 Eysenck's personality theory and organizational psychology 471

susceptible to stress are likely to perform less well than those who are more
stable. This association prompts a number of questions: whether all aviators
have this personality pattern; whether military pilots are preselected with
respect to it; or whether military flying training regimes filter out those who do
not have it.
More recently Bartram (1995) obtained 16PF and EPI data from 528 male
army applicants for flying training. He found the applicants highly "self-
selected" being much more emotionally stable and more extraverted than the
general population. He found both questionnaires predictive, with single
validities in the .20 to .30 range. He concluded:
In conclusion, the results of this study provide further support for the role of
personality measures in predicting flying training outcome. While the effects found
are relatively small, they are consistent with expectations and earlier research. Even
quite small increments in validity (or rs = .10-.20) will result in very substantial cost-
benefit in flying training—and in subsequent operational flying. Personality variance
is relatively independent of that which is otherwise assessed during selection
(primarily ability and motivation). As a result, measures of personality can potentially
yield useful increases in the overall validity of the selection process for flying training,
(pp. 234-235)
Second, a good deal of research on the differences between police officers
and the general population relies on normative data for its comparisons. Potter
and Cook's work (1977, reported in Colman and Gorman, 1982), and
described in a study by Clucas (Colman & Gorman, 1982), found 219 police
officers in a northern British police force to be relatively extraverted, tough-
minded, and conservative compared to normative data tested by Eysenck's
Social and Political Attitude Inventory, Eysenck's Personality Inventory, and
some other tests. Gudjonsson and Adlar (1983), working with eight British
police forces, administered Eysenck's Personality Questionnaire (which
measures extraversion, neuroticism, psychoticism, and lying) and Eysenck's
15 (which measures impulsiveness, venturesomeness, and empathy) to 84
recruits, 84 probationary constables with about 18 months' experience, 73
experienced constables with a mean length of service of 19.9 years, and 112
officers of the rank of inspector or above with a mean length of service of 19.5
years. The recruits were highly extraverted, impulsive, and venturesome when
their scores were compared to normative data. Their empathy scores did not
differ from the normative scores, but the empathy scores for the other three
groups were significantly lower than the norm. The experienced constables and
senior officers were significantly less impulsive and venturesome than the norm
and the recruits. The probationers, experienced constables, and senior officers
did not differ from the norm for extraversion. Unfortunately (unlike the EPQ
data), the normative data for Eysenck's 15 is not scaled for age, so the changing
results could be an artifact of age, rather than the result of socialization, as one
472 Further Eysenckian interests

would expect to find decreasing impulsiveness and venturesomeness with

increasing age. However, the reducing extraversion and empathy scores seem
likely to be caused by socialization, as they occur by 18 months' service.
Burbeck and Furnham (1984) found that 319 applicants to the Metropolitan
police displayed much higher levels of extravert stability on Eysenck's
Personality Inventory than the norm. These results were obtained before the
candidates were selected and included those who were turned down, which
would be evidence for the predispositional model. However, because the test
was administered at the time of the selection interview, the authors had reason
to believe that there was extensive "faking good" on the part of the applicants.
Finally, what happens if an individual finds himself/herself in a job that is ill-
suited to his personality? If he remains in that job, then the obvious answer is
that his job performance will tend to be relatively poor. An alternative
possibility that has rarely been considered is that his personality may alter as a
result of being exposed to a particular job environment. Turnbull (1976) found
that there was no tendency for success among male student salesmen to be
related to extraversion. However, the experience of selling and making
numerous contacts with strangers produced a highly significant increase in the
average level of extraversion.
It should be noted that nearly all the work-specific questionnaires designed
primarily for job selection tap into the fundamental Eysenckian dimension
(PEN) though they do not always use his terminology.

4. PERSONALITY AND ACCIDENTS

The Eysenckian trait dimensions have also been found to predict "negative"
occupational variables. There has been some interest in the relevance of
personality to performance under rather monotonous conditions. It might be
predicted that underaroused extraverts would find it more difficult than
introverts to maintain performance over time. Extraverts showed a greater
deterioration than introverts in driving performance over a 4-h period
(Fagerstrom & Lisper, 1977). However, their performance improved more
than that of the introverts when someone talked to them or the car radio was
turned on.
Shaw and Sichel (1970) compared the personality characteristics of accident-
prone and safe South African bus drivers (see Figure 21.1). Most of the
accident-prone drivers were neurotic extraverts, whereas the safe drivers were
predominately stable introverts. As might have been expected, it is the
impulsiveness component of extraversion rather than the sociability
component that is more closely related to poorer driving and accident
proneness (Loo, 1979).
 Eysenck's personality theory and organizational psychology 473

Neurotic
Onvtrs with bod
x Occident records

Introvert- Extrovert
0
o
0
o o

Safe drivers
o
Stable
Figure 21.1. Personality differences between accident-prone and safe bus drivers. Source: Shaw
and Sichel (1970).

There are a number of studies examining the relationship between person-

ality (particularly extraversion and neuroticism) and accidents (particularly
motor car accidents) (Furnham & Saipe, 1993). Despite various methodol-
ogical difficulties and differences, the results are fairly consistent. Venables
(1956) found, as predicted, that driver consistency is related to neuroticism and
extraversion in some groups. Presumably, erratic and inconsistent behavior is
closely associated with accidents.
Fine (1963) argued the following: since extraverts are assumed to be less
socialized than introverts, it is reasonable to assume that they should be less
bound by prescribed rules of society regarding motor vehicle operation.
Therefore, it was hypothesized that they would incur more traffic accidents and
violations than introverts (pp. 95-96). The study done on 937 college students
showed, as predicted, that extraverts had more accidents and traffic violations
than introverts.
Similarly, Craske (1968), who investigated 70 men and 30 women being
treated after accidents in a minor trauma clinic, found a highly significant
correlation between extraversion and accidents among the men, but not the
women. Moreover, the positive correlation between accidents and extraversion
was not significantly altered by examining severe and minor accidents in detail.
A closer examination of the actual test items related to accidents showed that
the few extraversion questions significantly related to accident repetition
tended to be related to impulsiveness rather than sociability, while the three
neuroticism questions were all concerned with guilt or depression. Schenk and
Rausche (1979) also found neuroticism closely associated with driving
accidents.
474 Further Eysencldan interests

In a more comprehensive, multivariate study, Hansen (1989) looked at

biodata, personality, and cognitive correlates of accidents in a causal model.
Looking at 362 chemical industry workers, it was hypothesized that traits of
social maladjustment, various aspects of neurosis, cognitive ability, employee
age, and actual job experience would have independent causal effects on
accidents, even when risk was partialled out. A social maladjustment scale was
constructed from the MMPI, which along with a measure of (neurotic)
distractibility was clearly linked with accidents. Both scales, though correlated,
demonstrated independent causal relationships to accidents suggesting two
major factors at work. Hansen believes that the central psychological question
of psychologists should be changed from "What personality or cognitive trait is
related to accidents?" to "What is the strength of the causal impact that trait
anxiety has on accidents?"
Finally, Booysen and Erasmus (1989) have done a Stirling job in reviewing
personality factors associated with accident risk. No less than 43 traits (many of
them related) have been examined as regards their relationship to accidents. In
a conceptual factor analysis, they suggested that two factors were relevant:
recklessness (extraversion, domineering, aggressive, sensation-seeking) and
anxiety-depressive. They then admitted the 16PF to nearly 200 bus drivers who
were divided into three groups depending on their previous involvement in
accidents and the degree of seriousness of accidents that they had been
involved in. A step-wise multiple regression showed that four factors—
dominance, carefreeness, emotional sensitivity, and shrewdness—accounted
for between 10 and 12% of the variance.
Thus, it seems that there is sufficient evidence that personality variables do
relate to all sorts of accidents in all sorts of populations. They appear to be able
to account for about 10% of the variance, which is certainly not to be
dismissed. The two orthogonal factors that seem to be the best predictors of
accidents are clearly extraversion/sensation-seeking/A-type behavior and
neuroticism/anxiety/instability; which is, of course, clearly in accordance with
Eysenck's (1967) theory.

5. PERSONALITY AND TRAINING

Many organizations invest a great deal of money in developing various skills

through training. Employees at all levels are sent on courses varying in length,
topic, and teaching style in order to ensure that they can perform more
efficiently and effectively. Apart from the more difficult questions to answer,
there is a central problem concerning whether training works. A major issue
concerns individual differences in learning styles. Kolb (1976) and others have
argued that people have preferred quite different learning styles. Presumably,
people learn more efficiently and effectively when taught in their preferred
 Eysenck's personality theory and organizational psychology 475

way. However, the evidence linking personality and preferred learning style is
not clear. There is currently a move to develop specialist questionnaires to look
at things such as learning styles. However, recent studies suggest the overlap
between these measures and the classic Eysenckian dimensions is such that
they add no new incremental validity (Furnham, 1992b).
Recent studies have shown that aptitude batteries predict success in training
moderately well. For instance, Dunbar and Novick (1988) found that abilities
like arithmetic reasoning and word knowledge were predictive of nine quite
different training criteria, although they did find considerable evidence of sex
differences.
There is, however, an extensive literature in cognitive psychology showing
how individual differences in such things as neuroticism and extraversion are
related to learning through differences in verbal learning, memory, and
performance. For instance, Eysenck (1981) has noted the following:
In spite of the relatively small volume of research on the effects of introversion-
extraversion on learning and memory, there appear to be a number of fairly robust
findings. Some of the more important of these have been discussed earlier and will
now be listed:
1) Reward enhances the performance of extraverts more than introverts, whereas
punishment impairs the performance of introverts more than extraverts.
2) Introverts are more susceptible than extraverts to distraction.
3) Introverts are more affected than extraverts by response competition.
4) Introverts take longer than extraverts to retrieve information from long-term or
permanent storage, especially non-dominant information.
5) Introverts have higher response criteria than extraverts.
6) Extraverts show better retention-test performance than introverts at short
retention intervals, but the opposite happens at long retention intervals.
While it is probably premature to attempt any theoretical integration of these various
findings, it is nevertheless tempting to argue that introverts are characteristically
better motivated on performance tasks than extraverts, with the consequence that
their normal expenditure of effort and utilization of working memory capacity is
closer to the maximum. Since introverts, as it were, start from a high motivational
baseline, it follows that they are less able than extraverts to utilize extra processing
resources to handle increasing processing demands (e.g. from distracting stimulation,
from response competition or from difficult retrieval tasks), (pp. 203-204)
There is also a small but interesting and important literature on the
interaction between personality and teaching methods. For instance Leith
(1972) demonstrated that extraverts learn much better than introverts with the
discovery method of learning, while introverts learn much better than
extraverts with the direct (reception) method.
Eysenck (1978) has listed six practical applications for personality variables
to learning situations: selection (advice based on personality traits as to fit);
streaming and setting (streaming pupils/students by personality or setting the
476 Further Eysenckian interests

different tasks); re-education (intervention in learning difficulties based on the

understanding of traits); ascertainment (monitoring of personality develop-
ment over time to anticipate problems); training (the education of teachers in
differential psychology); and research which takes seriously the role of
individual differences.
Furnham and Medhurst (1995) found the EPQ was a fairly good predictor of
student behavior in seminars (see Table 21.2). Psychoticism was a consistent
and powerful predictor of negative seminar behavior and academic
performance, while the Lie scale showed the opposite pattern. Extraversion
was, predictably, correlated with participation in the seminar and both
predicted and actually obtained poorer final exam results, though not with
essays handed in. There was only one significant correlate of neuroticism. High
scorers were predicted to do less well than stable individuals.
The results provide modest support for Eysenck's theory. However, what the
results of this study do highlight was that rather than the two most fundamental
and debated personality variables of extraversion and neuroticism being
predictors of seminar behavior and academic performance it was psychoticism
(tough-mindedness) and the Lie scale that were the most consistent predictors.
Despite the skewdness in the P scale it seemed to be a very predictable
correlate of poor performance. Equally the Lie scale, which is an index of
social conformity, seemed a rather good predictor of both seminar
performance and final year marks. It is quite possible that similar results
may be found from studies on occupational training courses.

6. NEUROTICISM AND JOB SATISFACTION

A number of studies have suggested that neurotics are less productive and
satisfied than non-neurotics. In a natural experiment, Organ (1975b) observed
Table 21.2. Eysenckian correlation of seminar reports (N = 72)

X Extraversion Neuroticism Psychoticism Lie

(A) Grasp of subject matter 0.75 -0.06 -0.04 -0.23* -0.28**
(B) Work habits 0.78 -0.12 -0.03 -0.38* 0.19*
(C) Motivation 0.75 -0.10 0.02 -0.40* * 0.26**
(D) Written expression 0.72 -0.09 0.03 -0.28* 0.26**
(E) Oral expression 0.69 0.08 -0.04 -0.28* 0.23*
(F) Participation in seminar 0.66 0.17* -0.06 -0.23* 0.12
(G) Estimate of performance8 0.68 0.15* 0.15* 0.39* * -0.25*
(H) Attendance 0.62 -0.10 -0.03 -0.25* 0.05
(I) Essays 0.62 -0.15* 0.11 -0.31* 0.15*
(J) Actual final degreeb 0.15* 0.01 0.24* -0.15*
a
High scores indicate low performance (1st & 3rd).
b
This is based on a N = 117 and high scores indicate low performance.
*p < 0.05; **p < 0.01; ***p <0.001.
 Eysenck's personality theory and organizational psychology 477

Table 21.3. Correlation between the four personality measures and the eight job satisfaction
factors (N = 88)

Personality scale

Job satisfaction factors Psychoticism Extraversion Neuroticism Lie

1. Supervision -0.17* 0.01 -0.12 0.15
2. Nature of the work -0.21* -0.11 -0.14 0.32***
3. Amount of work -0.01 0.10 -0.33*** 0.28**
4. Working conditions -0.05 0.10 -0.08 0.10
5. Co-workers -0.19* 0.05 -0.31*** 0.21*
6. Pay 0.01 0.20* -0.29** 0.36***
7. Future with the organization -0.007 0.04 -0.01 0.01
8. Overall job satisfaction -0.03 0.18* -0.06 0.10

*p < 0.05; **/? < 0.01; **/> < 0.001.

Source: Furnham and Zacherl (1986).

neurotic and non-neurotic business students as they took structured or "ambi-

guous" exams under high or low pressure. Predictably, the neurotics reported
much more emotional stress in the ambiguous exam than non-neurotics.
Furnham and Zacherl (1986) examined the relationship between personality
and job satisfaction as measured by a multidimensional scale. The results are
shown in Table 21.3. Both the psychoticism and neuroticism scales are
correlated negatively with all the subscale scores while both extraversion and
the lie scale correlated positively with all of the subscale scores. People with
high psychoticism scores (tough-minded) tended to be significantly less
satisfied with their supervisors, the nature of the work and their co-workers
than people with low psychoticism scores (tender-minded).
People with high neuroticism scores (unstable neurotics) tended to be fairly
highly significantly less satisfied with the amount of work that they were
required to do, their co-workers, and their pay. But it was the lie scale—a
measure of desirability—which in fact yielded the most and the biggest
correlations. The fact that neuroticism correlates consistently negatively with
the job satisfaction factors suggests that neurotics are in general less job
satisfied than non-neurotics. This may be because they are less productive, or
rather that their poor satisfaction might lead to poorer productivity. Whichever
way the direction of causality, it seems to be that neurotics make less satisfied
employees than non-neurotics.
Levin and Stokes (1989) looked at the trait of "negative affectivity" (NA)
which for them is a mix of anxiety, irritability, neuroticism, and self-
depreciation. They argue:
High-NA individuals have ongoing feelings of distress and nervousness. They tend to
dwell on their mistakes, disappointments, and shortcomings and to focus more on the
negative aspects of the world in general. In contrast, low-NA individuals appear to be
more satisfied, self-secure, and calm and to focus less on, and be more resilient in
response to life's daily frustrations and irritations, (p. 753)
478 Further Eysencfaan interests

In both a laboratory and a "natural" experiment, they found that negative

affectivity (neuroticism) was related to lower job satisfaction. They argue that
non-neurotics may be denying or repressing various frustrations, disappoint-
ments, and problems, or that the cognitive processes of neurotics lead them to
perceive the world more negatively. Organizations would seem wise to screen
out extreme neurotics and those with negative affectivity.
Perone, De Waard, and Baron (1979) found similar correlations when
examining satisfaction with real and stimulated jobs. They found neuroticism
and sensation seeking were negatively correlated with satisfaction, indicating
that dissatisfaction is symptomatic of general emotional maladjustment.
Thus, it seems that neuroticism is a highly undesirable trait in the workplace.
Yet there is fairly consistent evidence that neuroticism is correlated with
academic success. McKenzie (1989) has reviewed two explanations for this:
First that neuroticism only correlates positively with success in highly selected
groups—particularly those selected for intelligence. Second, neuroticism only
correlates positively with achievement in people that have appropriate coping
strategies and super-ego strength. There is in fact evidence for both, suggesting
that if neuroticism is "moderated" by intelligence and appropriate coping
skills, it will not seriously inhibit achievement.

7. PERSONALITY AND DISTRACTIBILITY AT WORK

Eysenck (1967) has argued that introverts and extraverts differ in terms of their
cortical arousal. Those who are classified as introverts have been shown to have
a lower optimum arousal threshold and therefore do not need much
stimulation before passing their optimum functioning level. Those who are
extraverts have higher optimum arousal thresholds and therefore tend to seek
arousal or stimulating situations. Gray (1964) linked these categories with the
Russian ideas of strong (extravert) and weak (introvert) nervous systems.
Vermolayever-Tomina (1964) found that those with a strong nervous system
tended to learn more in distracting situations than those who have a weak
nervous system. This study therefore hypothesizes that introverts would be
more negatively, and extraverts more positively, affected by the introduction of
extra stimulation, for example music, into their work environment.
It has been demonstrated that when studying in a library, introverts were
significantly more likely to choose a place to work away from the bustle of
certain areas, while the extraverts were more attracted to the latter as a work
place (Campbell & Hawley, 1982). This provides further evidence of the
regulation of arousal differences between introverts and extraverts.
Morgenstein, Hodgson, and Law (1974) found that extraverts actually
performed better in the presence of distractions than they did in silence, while
introverts showed a deficit in performance. Their subjects were asked to attend
 Eysenck's personality theory and organizational psychology 479

to, and remember, a number of words out of a long list that was read to them,
whilst they were being read a passage by the same voice. They were given the
means of controlling the balance of sound between the word list and the
passage, but the greater this difference, the more words to be remembered
were distorted. The study posed three questions: Is the preference for
distortion or distraction related to the personality dimension of introversion/
extraversion? Do the two groups of subjects differ in their performance on the
task? How did the subject arrive at their preferred balance? They found that
extraverts make extravagant sweeping movements in their effort to find a
balance, while introverts make much fewer, smaller adjustments. This finding
was consistent with Eysenck's theory that the introvert's nervous system is
overdamped. There was a trend for introverts to avoid distraction when the
personality dimension was compared with choice of distortion/distraction, and
they did not perform the task as well although the effect was not statistically
significant. They concluded:
It would seem that the extraverted subjects do not merely prefer to be in the company
of others, but that their work efficiency actually improves in the face of distractions,
while the solitary preferences of the introverts are reflected in their reduced
efficiency of work when distracted. Paying heed to such preferences, as measured by
the Eysenck Personality Inventory, is therefore not only a method of increasing
contentment at work by means of personnel selection, but should also result in
improved efficiency of output, (p. 220)
Various studies have examined the distracting effects of television on
cognitive processing. Recent research on television distraction effects
(Armstrong & Greenberg, 1990; Armstrong, Boiarsky, & Mares, 1991)
reported significant performance decrements for several measures, that is,
spatial problem solving, mental flexibility, and reading comprehension as a
function of television. These results were consistent with the idea that
background television influences performance by causing cognitive processing
limits to be exceeded on complex tasks, while indicative of a television
distraction influence on parallel cognitive activities. Armstrong's research did
not investigate the possibility of individual differences among children in their
parallel processing capabilities. This point is particularly pertinent in the light
of psychological research showing that personality factors such as introversion-
extraversion are important mediators if individual cognitive performance in the
progress of distraction (Morgenstern, Hodgson, & Law, 1974). Furnham,
Gunter, and Peterson (1994) conducted a study into the effects of the presence
of an operating television on introverts and extraverts, while they completed
reading comprehension tasks. They found, as predicted, a significant inter-
action F(l,39) = 7.41 p < 0.01 between the personality dimension and the
treatment effect. In other words, the introverts and extraverts performed
equally well with the television off, but the extraverts performed better than the
introverts when the television was on.
480 Further Eysenckian interests

More recently Furnham and Bradley (1997) looked at the distracting effects
of "pop music" on introvert's and extravert's performances on various
cognitive tasks. It was predicted that there would be a main effect for music
and an interaction effect with introverts performing less well in the presence of
music. Introverts and extraverts were given two tests, a memory test (with
immediate and delayed recall) and a reading comprehension test, which were
completed either while being exposed to pop music, or in silence. The results
showed that there was a detrimental effect on immediate recall on the memory
test for both groups when music was played, but no main effect in the other
condition. However, two of the three interactions were significant. After a 6-
min interval the introverts who had memorized the objects hi the presence of
pop music had a significantly lower recall than the extraverts in the same
condition and also the introverts that had observed them in silence. The
introverts who completed a reading comprehension task when music was being
played also performed significantly lower than these two groups. These
findings have implications for the study habits of introverts when needing to
retain or process complex information.
The findings of these studies are relevant to all those who work in a
communal area, be it an open-plan office or a student work room. Some
people may thrive on background noises while others, the extreme introverts
will find it immensely debilitating. This consideration is important for
management who wish to optimize the output of their work-force, and
minimize the space they work in.

8. THE GENETIC DETERMINANTS OF JOB SATISFACTION

Eysenck has always been interested in the biosocial model of man and behavior
genetics. He has stated consistently that there is a significant amount of the
variance in IQ and personality scores that is directly attributable to inheritance.
But it is not until recently that behavior geneticists have seriously looked into
the possibility that work behaviors are genetically determined.
Arvey, Bouchard, Segal, and Abraham (1989) investigated 34 monozygotic
twins reared apart to look at genetic and environmental components in job
satisfaction. They were quite clearly provoked by the article by Staw and Ross
(1985) who noted:
Job attitudes may reflect a biologically based trait that predisposes individuals to see
positive or negative content in their lives ... Differences in individual temperament...
ranging from clinical depression to a very positive disposition, could influence the
information individuals input, recall and interpret within various social situations,
including work. (p. 471)
There is, they argue quite correctly, no reason to believe that genetic factors do not
 Eysenck's personality theory and organizational psychology 481

affect job satisfaction, and thus examined 34 monozygotic twins (reared apart) just
over 40-years-old on average. They completed a multidimensional questionnaire on
job satisfaction and the results showed a significant hereditability of intrinsic and
general, but not extrinsic and overall satisfaction. They also demonstrated clear
evidence that there is a genetic component in terms of the job that is sought and held
by individuals. However, the fact that the sample held similar jobs cannot account for
the hereditability coefficient being significant.
The authors note, however, that the genetic factor accounted for about 30%
of the variance, which is not overwhelming, but does not necessarily imply that
all of the remaining variance is due to the environment. The authors argue that
intellectual capacity (shown to have a strong hereditary component) probably
accounts for the similarity between the jobs chosen by the twins. They argue
that these results have two major implications. First, that workers bring
dispositions to jobs that are more difficult to modify than heretofore
acknowledged. Thus, job enrichment and other satisfaction-increasing
programs might miss the mean levels of satisfaction for workers, but the
rank order is preserved. Second, that future satisfaction may be predicted from
current satisfaction, that is, it can be used as a criterion for prediction.
These results will remain, like the whole issue, extremely contentious, not so
much because of the methods used but the sociopolitical implications of these
results for selection and, more importantly, organizational and structural
attempts to improve job satisfaction (and hence perhaps satisfaction). Hans
Eysenck would certainly express no surprise at these results. Furthermore he
would, no doubt, fully understand the implications of these findings for
attaining job satisfaction.

9. SENSITIVITY TO REWARDS AND PUNISHMENTS, AND JOB SATISFACTION

The motivational differences between extraverts who are motivated to seek

rewards and introverts who are motivated to avoid punishment have been
examined by Gray (1973) who, in rotating Eysenck's two factors along a 45 axis
(see chapter 3), has presented one of the most coherent challenges to that
theory. Gray's challenge to his former supervisor is well known but its
organizational implications less considered.
Gray's theory asserts that extraverts will respond more readily to reward
while introverts react primarily to punishment. Although extraverts will react
positively to an achieved reward and introverts will react positively to an
applied punishment, both extraverts and introverts perceive reward and
punishment in terms of possible current or future realizations. The extravert is
motivated to gain a promised reward; the introvert is motivated to avoid a
threatened punishment. Also, the overapplication of the principle tends to
lessen the intended effects, dampening the motivational qualities of reward
482 Further Eysenckian interests

and punishment: because the extravert is motivated by opportunity to gain

reward, too much rewarding reinforcement tends to create a satiating effect on
the extravert's desire to achieve. Also, since the introvert is motivated by a
need to avoid punishment, too many threats or actual enforcement of the
negative reinforcement places the introvert in the position of being unable to
avoid punishment, and so he/she becomes immobilized and the motivational
effect of punishment is decreased of course. The motivating effects of reward
and punishment are not mutually exclusive: an extravert does not wish to be
punished and will react to negative reinforcement, and all introverts want to be
rewarded and are motivated by positive reinforcement.
The two crucial factors in this theory are: first, the tendency of the extravert
or introvert to perform more satisfactorily in the face of either reward or
punishment. Second, the degree of extraversion or introversion in a given
personality. The more extraverted, the greater the sensitivity a person has
towards promises of reward, while a person closer to the introversion end
would display greater sensitivity towards threats of punishment.
The practical application of Gray's theory to occupational settings is
appealingly obvious, and we can now more effectively apply the carrot/stick
principle in socializing human behavioral responses. It becomes apparent that
it would be a waste of time to try to motivate an extravert with threats of dire
punishment (such as sacking, no pay rise), and it would prove equally
unsuccessful to attempt to entice an introvert with promise of pay and benefits.
To exact the highest level of performance from individuals, motivators must
encourage the extravert with potential rewards and prompt the introvert with
judicial use of punitive threats. Thus, extraverted organizations, like those
involved in selling, could best motivate and satisfy their staff by providing
regular but varied rewards. Equally, a primarily introverted organization, as in
many bureaucracies, can best shape or motivate staff by the threat of sanctions.
Gray's theory concerning sensitivity to signals of reward and punishment
attempts to explain individual differences in extraversion and introversion but
also deals with neuroticism. Just as extraversion and introversion can be viewed
on a continuum scale, so, too, can individuals be evaluated on a continuum of
stability and neuroticism.
The degree of neuroticism heightens an individual's sensitivity to reward or
punishment. The introvert, sensitive to punishment, who displays high
neuroticism becomes more sensitive to both reward and punishment with the
greatest increase being toward punishment. That is, the neurotic introvert
becomes more concerned with reward but is even more anxious about
punishment than the low neuroticism introvert. As neuroticism increases, the
extravert (sensitive to reward) becomes more sensitive to both reward and
punishment, with high increases in reward sensitivity. Although extraverts and
 Eysenck's personality theory and organizational psychology 483

introverts increase in sensitivity to reward and punishment as neuroticism

increases, each has the highest increase of sensitivity to that trait commonly
attributed to extraversion or introversion.
Thus, an extraverted neurotic, being highly sensitive to reward, is less
socializable in terms of legal, organizational norms and more likely to become
maladaptive or difficult. Given moderate levels of extraversion, high N
(neurotic) individuals are usually more responsive to control techniques than
low N (stable) individuals. Whether reward or punishment is the controlling
factor, the oversocialized individual will respond readily and may tend to
become overcontrolled, while undersocialized individuals may show little or no
response to control measures. Consequently, the low N (stable) individual may
necessitate the use of rigid control and severe disciplinary measures
(Wakefield, 1979).
Empirical support for this thesis has come from various sources. Gupta
(1976) used a linguistic task to show that extraverts condition more readily to
reinforcement and introverts to punishment. Gupta used two experimenters,
although all subjects were male, one experimenter was male and the other was
female. When the word "good" was vocalized by the female, the young male
extraverts showed a more significant response differential than when the word
was spoken by the male experimenter. Gupta concluded that, apparently, the
encouraging word from the female was sufficiently rewarding while it appeared
probable that the more explicit reward was required from the male. "The
strength of conditioning is to a certain extent determined by individual's
subjective attitude towards the person who administers the reinforcement"
(Gupta, 1976, p. 50).
Similarly, Wakefield and colleagues tried to apply Gray's theory to
educational settings. They argued that achievement in the elementary
classroom can be improved by applying differential modes of reinforcement
to extraverts and introverts. Extraverts should be rewarded with extensive
praise and consistently encouraged by reminders of potential rewards
commensurate with competent performance. Introverts, on the other hand,
should be judicially exposed to threats of punishment and made continually
aware of the negative sanctions resulting from unsatisfactory performance.
McCord and Wakefield (1981) hypothesized that: (1) introverts have better
arithmetic achievements than extraverts when exposed to higher levels of
teacher-presented punishment in the classroom, and (2) a reversal would occur
in which extraverts would achieve arithmetic advantages in classroom
situations where a teacher-presented reward was prevalent. They related the
reward to punishment ratio to teachers, the personality of children, and the
arithmetic performance of elementary school pupils. They found that
extraverts do meet expectations of higher achievement than introverts in
484 Further Eysenckian interests

classrooms where there is a predominance of teacher-presented reward, but

when the gap between reward and punishment predomination narrows,
introverts have a greater achievement.
Boddy, Carver, and Rowley (1986) gave introverts and extraverts two tasks
to do: a computer game involving initiation of cursor movements on a VDU to
find a hidden target, and a task involving receding of decimal numbers and
letters, and doing calculations. As predicted, extraverts performed better under
positive than negative reinforcement, while introverts performed better under
negative than positive reinforcement. In a study looking at reactions to
punishment, Patterson, Kosson, and Newman (1987) found extraverts fail to
pause following punishing errors, but that longer pausing following punish-
ment predicted better learning from punishment for both introverts and
extraverts. They note:
In the presence of reward incentive, extraverts are more prone to facilitate their
approach behaviour than to elicit interruption and reflectivity. Without adequate
reflection, extraverts fail to associate punishment with the incorrect response and are
therefore less likely to inhibit that response on subsequent occasions. In contrast to
stable extraverts, whose disinhibited reaction to punishment appears to depend on
the presence of cues for reward, the reaction of neurotic extraverts appears to be less
situationally determined. To the extent that this disinhibited reaction to punishment
interferes with learning and subsequent inhibition, we might expect that stable
extraverts' insensitivity to punishment will be more situation specific than neurotic
extraverts." (p. 570)
A number of attempts have been made to devise measures of Gray's theory.
For instance, Torrubia and Tobena (1984) devised a "susceptibility to
punishment" scale which showed predictable and satisfactory correlations
with Eysenck's measure. Wilson, Barrett, and Gray (1989) were less successful,
however. They devised a five-dimension measure—approach, active avoidance,
passive avoidance, extinction, and fight/flight—which, although they showed
satisfactory internal consistency, did not correlate with the Eysenckian
dimensions as hypothesized.
Given the nature of this theory, what are the implications for organizational
behavior? Furnham (1992a) has speculated that the organizational incentives
(i.e. performance related to pay, promotion possibilities) and prohibitions (i.e.
potential sacking, fining) work differently for different people in the
organization. Extravert organizations (that is, those that are dominated by
extraverts) can motivate and shape staff by having small (but incremental and
worthwhile) incentives that act as reinforcements. The more regular, consist-
ent, and public these are, the better. "Sales-person of the month" and annual
awards for efficiency, tact, customer relations, etc. are likely to be more
successful with extraverts. Introverted organizations (that is, those that are
dominated by introverts, and highly sensitive to potential sanctions and
punishments) could be used to shape, or at least prevent, various kinds of
 Eysenck's personality theory and organizational psychology 485

behaviors. Thus, threats of imminent job loss, compulsory retirement, working

on half-time are likely to make introverts work harder than extraverts.
Organizations dominated by extraverts would do well to maintain a "culture"
where people give each other open, honest, and regular positive feedback for
work well done, while introverted cultures would have ways to remind people
regularly that stepping out of line or underperforming will be punished. The
obvious major implication of this work is that management systems devised to
regulate the behavior of employees should be sensitive to major individual
differences. The carrot and the stick should both be available, but they will not
have equal effect on all employees.

10. THE EYSENCK PERSONALITY PROFILER (EPP)

The development of the Eysenck personality measures is well known to the

London School and his followers: MPI, EPI, EPQ, EPQ(R) and most recently
the EPP. Although the earlier measures, particularly the EPI and EPQ, have
been used extensively in organizational psychological research, the EPP has
been widely adopted in commercial organizations as a personality measure
useful in selection, training, succession, planning, etc.
The attraction for the commercial world of the EPP is that it measures not
only the famous super factors (PEN) but 21 primary factors (seven for each
super factor). Further these are described in every-day, less clinically oriented,
terminology. Thus one of the neuroticism factors is labeled "Dependent
Autonomous", one from extraversion is called "Ambitious-Unambitious" and
one from the psychoticism super factor is labeled "Risk-taking—Careful." The
possibility of having as many as 21 factors is highly appealing to personnel and
human resource directors who like "rich" descriptions of potential employees
about whom to make judgment.
The EPP has been described and critically appraised (Eysenck, Barrett,
Wilson, & Jackson, 1992). Costa and McCrae (1995) have provided
preliminary evidence supporting the convergent and discriminant validity, yet
they remain critical of the "misclassification" of some of the scales. Wilson and
Jackson (1994) have provided some construct validity evidence when they
demonstrated that physicists tend to be more introverted and cautious
(particularly careful, controlled, inhibited, and unsocial) than the general
population.

11. CONCLUSION

Whilst Eysenck has been less concerned in applying his theory and research to
organizational issues, there is ample evidence that his students and other
486 Further Eysenckian interests

followers have seen numerous obvious applications particularly of his

personality theory. Further, it is likely that his more recent work on health
and creativity will, in due course, be researched within the organizational
context.
Furnham (1992a) pointed out that there has, to date, been a poor
rapprochement between personality theorists and occupational psychologists.
The basic tenet of this "classic personality theory" approach is to measure
personality as the independent variable and to see how it correlates with some
(often rather arbitrarily chosen) work-related behavior. But the approach has
been piecemeal and there is very little evidence of a concerted, systematic, and
programmatic research effort, which is perhaps not that unusual. Sometimes
this research has been laboratory based and hence it frequently has poor
ecological validity. Further the selection of work-related variables is somewhat
random and based on convenience, because researchers are either unable to
get better measures or, indeed, are not sure what to look for. Essentially,
studies such as these are nearly always seen by personality researchers simply
as supporting evidence for their ideas. Compared to the extensive research on
the relationship between personality and, say, learning, mental health, or social
behavior, the extant research from classic personality theory on occupational/
organizational variables has been disappointing.
On the other hand, researchers in organizational psychology are usually
interested in examining personality correlates of specific work behaviors which
might help personnel and human-resource professionals select, appraise,
promote, or train individuals. This research has a number of limitations: First,
the choice of personality variables has been arbitrary and uninformed. Some
personality tests have been favored mainly because they have been
commercially exploited rather than because they are reliable and valid. Some
outdated tests, largely forgotten and condemned by psychometricians, remain
a popular choice and hence seriously threaten the nature of the results.
Second, statistical analyses have been simple and naive. As a rule, simple
correlations have been computed rather than partial correlations, or even more
preferably multivariate statistics to prevent type II errors (finding more
significant differences than actually occur). Given that both independent and
dependent variables are multifactorial, it is essential that sufficiently robust
and sensitive multivariate statistics are used to analyze results. Third, studies in
this area are frequently exploratory and atheoretical rather than based on a
sound theory or programmatic research endeavor. As a result, interesting
results are rarely followed up and the theoretical implications rarely exploited.
Finally, researchers often ignore possible organizational and societal factors
that either directly or indirectly affect the dependent variable. That is, work-
related behaviors are rarely solely under the control of the individual and may
be moderated by powerful organizational factors which need to be taken into
account.
 Eysenck's personality theory and organizational psychology 487

The occupational psychology/organizational behavior literature is diverse,

often poor but sometimes very good. Alas, good research and theorizing is
difficult to find and limited in both quantity and scope. Certainly Hans
Eysenck's theories and research provide the opportunity and the incentive for
these two research traditions to benefit from each other.

REFERENCES

Armstrong, G, & Greenberg, B. (1990). Background television as an inhibitor of

cognitive processing. Human Communication Research, 16, 355-386.
Armstrong, C, Boiarsky, G., & Mares, M. (1991). Background Television and Reading
Performance, 58, 235-253.
Arvey, R., Bouchard, T., Segal, N., & Abraham, L. (1989). Job satisfaction:
Environmental and genetic components. Journal of Applied Psychology, 74, 187-192.
Barrett, P., Kline, P., Paltiel, L., & Eysenck, H. (1996). An evaluation of the psycho-
metric properties of the concepts 5.2 Occupational Personality Questionnaire.
Journal of Occupational and Organizational Psychology, 69, 1-19.
Bartram, D. (1995). The predictive validity of the EPI and 16PF for military flying
training. Journal of Occupational and Organizational Psychology, 68, 219-230.
Bartram, D. & Dale, H. (1982). The Eysenck Personality Inventory as a selection test
for military pilots. Journal of Occupational Psychology, 55, 287-296.
Bendig, A. (1963). The relationship of temperament traits of social extroversion and
emotionality to vocational interests. Journal of General Psychology, 69, 311-313.
Bishop, D. & Jean Renaud, G. (1976). End-of-day mood on work and leisure days in
relation to extraversion, neuroticism, and amount of change in daily activity.
Canadian Journal of Behavioural Science, 8, 388-400.
Blunt, P. (1978). Personality characteristics of a group of white South African managers.
Internal Journal of Psychology, 13, 139-146.
Boddy, J., Carver, A, & Rowley, K. (1986). Effects of positive and negative verbal
reinforcement on performance as a function of extraversion-introversion. Personality
and Individual Differences, 7, 81-88.
Booysen, A. & Erasmus, J. (1989). Die verband tissen enkele persoonuk-hendfaktore en
belsing risiks. South African Journal of Psychology, 55, 321-329.
Burbeck, E. & Furnham, A. (1984). Personality and police selection: Trait differences in
successful and non-successful applicants to the Metropolitan police. Personality and
Individual Differences, 5, 257-263.
Campbell, J. B. & Hawley, C. W. (1982). Study habits and Eysenck's theory of
extraversion-introversion. Journal in Research in Personality, 16, 139-146.
Christie, M. & Venables, P. (1973). Mood changes in reaction to age, EPI scores, time
of day. British Journal of Social and Clinical Psychology, 12, 61-72.
Claridge, G. (1986). Eysenck's contribution to the psychology of personality. In S.
Modgil & C. Modgil (Eds.), Hans Eysenck: Consensus and controversy (pp. 73-85).
London: Palmer Press.
Cohen, D. (1977). Psychologists on psychology. London: Routledge & Kegan Paul.
488 Further Eysenckian interests

Colman, A. & Gorman, L. (1982). Conservation, dogmatism and authoritarianism in

British police officers. Sociology, 16, 1-11.
Cook, M. (1985). Personnel selection and productivity. Chichester, U.K.: Wiley.
Cooper, R. & Payne, R. (1967). Extraversion and some aspects of work behaviour.
Personnel Psychology, 20, 45-47.
Costa, P. & McCrae, R. (1995). Primary traits of Eysenck's P-E-N system: Three- and
five-factor solutions. Journal of Personality and Social Psychology, 1995, 69 308-317.
Craske, S. (1968). A study of the relation between personality and accident history.
British Journal of Medical Psychology, 41, 399-404.
Dunbar, S. & Novick, M. (1988). On predicting success in training for men and women.
Journal of Applied Psychology, 73, 545-550.
Eysenck, H. (1953). Uses and abuses of psychology. Harmondsworth, U.K.: Penguin.
Eysenck, H. (1967a). The biological basis of personality. Springfield, IL.: Thomas.
Eysenck, H. (1967b). Personality patterns in various groups. Occupational Psychology,
41, 249-250.
Eysenck, H. (1971). Readings in extraversion-introversion II. Fields of Application.
London: Staples.
Eysenck, H. (1978). The development of personality and its relation to learning. In
Murray-Smith, S. (Ed.), Melbourne studies in education (pp. 134-181). Melbourne
University Press.
Eysenck, H. (1990). Rebel with a cause. London: W. H. Allen.
Eysenck, H. & Eysenck, M. (1985). Personality and individual differences: A natural
science approach. London: Plenum.
Eysenck, H., Barrett, P., Wilson, G., & Jackson, C. (1992). Primary traits measurement
of the 21 components of the P-E-N system. European Journal of Psychological
Assessment, 8, 109-117.
Eysenck, M. (1981). Learning, Memory and Personality. In Eysenck, H. (Ed.), A model
for personality. Heidelberg, Germany: Springer, pp. 169-207.
Fagerstrom, K. & Lisper, H. (1977). Effects of listening to car radio, experience, and
personality of the driver on a subsidiary reaction time and heart-rate in a long-term
driving task. In R. Machie (Ed.), Vigilance. New York: Plenum Press.
Fine, B. (1963). Introversion-extraversion and motor vehicle driver behavior. Perceptual
and Motor Skills, 12, 95-100.
Furnham, A. (1992a). Personality at Work. London: Routledge.
Furnham, A. (1992b). Personality and learning style. Personality and Individual
Differences, 13, 429-438.
Furnham, A. & Zacherl, M. (1986). Personality and job satisfaction. Personality and
Individual Differences, 7, 453-455.
Furnham, A. & Saipe, J. (1993). Personality correlates of convicted drivers. Personality
and Individual Differences, 14, 329-336.
Furnham, A., Gunter, B., & Peterson, E. (1994). Television distraction and the perfor-
mance of introverts and extraverts. Applied Cognitive Psychology, 8, 705-711.
Furnham, A. & Medhurst, S. (1995). Personality correlates of academic seminar
behaviour: A study of four instruments. Personality and Individual Differences, 19,
198-208.
 Eysenck's personality theory and organizational psychology 489

Furnham, A. & Bradley, A. (in press). Music while you work: The differential
distraction of background music on the cognitive test of performance of introverts
and extraverts. Applied Cognitive Psychology, 11.
Gibson, H. (1981). Hans Eysenck: The man and his work. London: Peter Owen.
Gray, J. (1964). Strength of the nervous system and levels of arousal: A reinterpretation. In
J. Gray (Ed.), Pavlov's typology (pp. 289-366). Oxford: Pergamon Press.
Gray, J. (1973). Causal theories of personality and how to test them. In J. Royce (Ed.),
Multivariate analysis and psychological theory. New York: Academic.
Gudjonsson, E. & Adlar, K. (1983). Personality patterns of British police officers.
Personality and Individual Differences, 4, 507-512.
Gupta, B. (1976). Extraversion and reinforcement in verbal operant conditioning.
British Journal of Psychology, 67, 47-52.
Hansen, C. (1989). A causal model of the relationship among accidents, biodata,
personality and cognitive factors. Journal of Applied Psychology, 74, 81-90.
Hill, A. (1975). Extraversion and variety-seeking in a monotonous task. British Journal
of Psychology, 66, 9-13.
Jessup, G. & Jessup, H. (1971). Validity of Eysenck's Personality Inventory in pilot
selection. Occupational Psychology, 45, 111-123.
Kim, J. (1980). Relationships of personality to perceptual and behavioural responses in
stimulating and non stimulating tasks. Academy of Management Journal, 23, 307-319.
Kirton, M. & Mulligan, G. (1973). Correlates of managers' attitudes toward change.
Journal of Applied Psychology, 58, 101-107.
Kolb, D. (1976). Learning style inventory: Technical manual. Boston, MA: McBer.
Leith, G. (1972). The relationship between intelligence, personality and creativity under
two conditions of stress. British Journal of Educational Psychology, 42, 240-247.
Levin, I. & Stokes, J. (1989). Dispositional approach to job satisfaction: Role of negative
affectivity. Journal of Applied Psychology, 74, 752-758.
Loo, R. (1979). Role of primary personality factors in the perception of traffic signs and
driver violations and accidents. Accident Analysis and Prevention, 11, 125-127.
McCord, R. & Wakefield, J. (1981). Arithmetic achievement as a function of
introversion-extraversion and teacher-presented reward and punishment. Personality
and Individual Differences, 2, 145-152.
McKenzie, J. (1989). Neuroticism and academic achievement: The Furneaux factor.
Personality and Individual Differences, 10, 509-515.
Modgil, S. & Modgil, C. (1986). Hans Eysenck: Consensus and controversy. London:
Falmer Press.
Morgenstern, S., Hodgson, R. J., & Law, L. (1974). Work efficiency and personality
ergonomics, 17, 211-220.
Okaue, M., Nakamura, M., & Niura, K. (1979). Satisfaction with real and simulated jobs
in relation to personality variables and drug use. Journal of Applied Psychology, 64,
660-668.
Organ, D. (1975a). Extraversion, locus of control, and individual differences in
conditionability in organizations. Journal of Applied Psychology, 60, 401-404.
Organ, D. (1975b). Effects of pressure and individual neuroticism on emotional
responses to task-role ambiguity. Journal of Applied Psychology, 60, 397-400.
490 Further Eysenckian interests

Patterson, C. Kosson, D., & Newman, J. (1987). Reaction to punishment, reflectivity,

and passive avoidance learning in extraverts. Journal of Personality and Social
Psychology, 52, 565-575.
Perone, M., De Waard, R., & Baron, A. (1979). Satisfaction with real and simulated
jobs in relation to personality variables and drug use. Journal of Applied Psychology,
64, 660-668.
Reinhardt, R. (1970). The outstanding jet pilot. American Journal of Psychiatry, 127,
732-736.
Rim, Y. (1961). Dimension of job incentives and personality. Acta Psychologica, 18,
332-336.
Rim, Y. (1977). Significance of work and personality. Journal of Occupational
Psychology, 50, 135-138.
Savage, R. & Stewart, R. (1972). Personality and the success of card-punch operators in
training. British Journal of Psychology, 63, 445-450.
Schenk, J. & Rausche, A. (1979). The personality of accident prone drivers. Psychologic
und Poaxis, 23, 241-247.
Shaw, L. & Sichel, H. (1970). Accident proneness. Oxford: Pergamon.
Staw, B. & Ross, J. (1985). Stability in the midst of change: A dispositional approach to
job attitudes. Journal of Applied Psychology, 70, 469-480.
Stern, L., Alexander, R., Barrett, G., & Dambrot, F. (1983). The relationship of
extraversion and neuroticism with job preferences and job satisfaction for clerical
employees. Journal of Occupational Psychology, 56, 145-155.
Torrubia, R. & Tobera, A. (1984). A scale for the assessment of 'susceptibility to
punishment' as a measure of anxiety. Personality and Individual Differences, 5, 371-
375.
Turnbull, A. (1976). Selling and the salesman: Prediction of success and personality
change. Psychological Reports, 38, 1175-1180.
Venables, P. (1956). Car driving consistency and measures of personality. Journal of
Applied Psychology, 40, 21-24.
Vermonlayeva-Tomina, L. B. (1964). Pavlov's typology. Oxford: Pergamon.
Wakefield, J. (1979). Using personality to individualize instruction. San Francisco, CA:
EDITS.
Wankowski, J. (1973). Temperament, motivation and academic achievement. University
of Birmingham: Unpublished report.
Wilson, G., Tunstall, O., & Eysenck, H. (1972). Measurement of motivation in
predicting industrial performance: A study of apprentice gas fitter. Occupational
Psychology, 46, 15-24.
Wilson, G., Barrett, P., & Gray, J. (1989). Human reactions to reward and punishment:
A questionnaire examination of Gray's personality theory. British Journal of
Psychology, 80, 509-515.
Wilson, G. & Jackson, C. (1994). The personality of physicists. Personality and
Individual Differences, 16, 187-189.
Chapter 22
Bursts of creativity and aberrant sunspot cycles:
hypothetical covariations*

S. Ertel

1. INTRODUCTION

Psychology, as a discipline, does not provide a conceptual framework for the

present chapter; its moving out of bounds requires justification. This chapter
approaches an exotic area, but here the author came upon H. J. Eysenck, who
has felt free to take up challenging problems whenever they could be subjected
to analytical-statistical methods. Eysenck disregarded, when necessary, the
discipline's predilections for conventional fields of inquiry.
After setting out to test Michel and Francoise Gauquelins' claim that there
exist connections between planetary positions and birth frequencies of eminent
professionals (see also chapter 23), to my own discomfort, I obtained evidence
in favor of this tenet which many in the sciences find unpalatable. Being a
member of my discipline and committed to its norms, I felt I should have been
almost ashamed at reaching such a conclusion, but to my surprise and relief I
realized that one of the most influential psychologists of our times had already
voiced views resembling my own:
Michel and Francoise Gauquelin ... provided what is perhaps the best evidence for
the truth of the hypothesis that the positions of the planets at birth have an influence
on the behaviour of human beings ... (Eysenck, 1978, p. 10)
I think it may be said that, as far as objectivity of observation, statistical significance
of differences, verification of the hypothesis, and replicability are concerned, there
are few sets of data in psychology which could compete with these observations.
(Eysenck, 1975, p. 249)
On my invitation, Professor Eysenck made a stopover at our Institute at
Gottingen and he agreed to be interviewed, in an open seminar, about his
interest in frontier science, an interview which was tape-recorded and

*Part of this chapter is based on a paper presented at the 26th International Congress of
Psychology in Montreal, 16-21 August 1996.
492 Further Eysenckian interests

published later (Ertel, 1986). The interviewer asked Professor Eysenck where
his interest in astrology and parapsychology, apparent antipodes of ordinary
psychological specialization, actually came from and how he could bear the
dissonance. Eysenck said:
Fundamentally, I was trained as a metrologist... My interest has always been from the
first in those areas where measurement is particularly difficult ... Problems of
parapsychology and of astrology are in the first instance problems of measurement,
(p. 136)
He rejected the interviewer's insinuation that his activity in these areas might
manifest some subconscious desire to give vent to suppressed occult beliefs.
No, I don't think so. I'm really not attracted by this, you know, I'm rather repelled by
it and I wish it were untrue. I'd be much happier if there were no parapsychological
phenomena and if there was no Gauquelin effect, but I can't deny on the evidence
that they do exist. So we have to admit them, and it therefore behooves us to look at
them. But I'm certainly not attracted by them in any way. No, it rather upsets me.
(p. 139)
Aversion is a scientist's prime reaction to fundamentally unexplainable
evidence. Such evidence will therefore be ignored by most, while a hard-core
minority may even reject, attack, or distort it (Benski et al., 1996; Ertel &
Irving, 1996). The number of those who will be stirred by the evidence to
impartial exploratory action is small. Eysenck was the first in our community
who approached the problem of possible connections between man and
extraterrestrial factors by suggesting research. The present chapter would not
have appeared in this book without his continuous encouragement.
In Astrology, Science and Superstition (1982, with Nias) Eysenck ventured an
excursion into heliobiology, a field of inquiry that might provide some missing
link for a better understanding of Gauquelin's planet-birth correlations: solar
activity, solar flares, solar wind, interplanetary field reversals, ionospheric
disturbance, geomagnetic perturbations, electromagnetic osculations in the
earth's atmosphere—any of these might become a bridge over the gap between
planetary positions as "stimuli" and the child's delivery as "response." By
getting hold of some such mechanism, those observations would begin to make
sense in scientific contexts.
The pioneer of heliobiological research was Alexander L. Chizhevsky (1924),
who claimed to have found strong indications of an "influence of cosmic
factors upon the behaviour of organized human masses." He even alleged that
some of these effects manifested themselves as social and political revolutions.
Chizhevsky's bombastic claim upset me. When I searched the literature, I did
not find any serious empirical attempt to disprove this claim. I then checked
Chizhevsky's data and method myself. I found bias and errors, but on
collecting new data, I actually obtained more support for his claims than
 Bursts of creativity 493

counter-evidence. Professor Eysenck, who had come upon this track on his
own, perused a preliminary report of these studies on his 1986 visit to
Gottingen, and his reaction to it raised the author's spirits.
Unlike research into areas related to the Gauquelin effects, research in the
Chizhevsky tradition has gained ground among a courageous minority of
researchers in various disciplines, including astronomers, geophysicists,
biogeophysicists, magnetobiologists, biometereologists, chronobiologists, and
chronomedical researchers. Not so among psychologists, however, even though
their methodological, and especially statistical, know-how provides a sound
basis for valuable contributions in this area. The findings presented below need
to be checked by metrologists, to use Eysenck's term, abounding in our
departments. Above all, replications using independent data are required.
Such data is easy to collect, and some sources may even be found on the
Internet. Thus, the author's hypotheses, if untrue, can be refuted without much
effort
That is the way in which science should be done ... Anybody can be mistaken. There
may be sources of errors nobody has thought of ... (Eysenck, 1986, p. 145)
So the only way you can possibly see whether there is anything to it is replication: Do
the whole thing again and again and again. If you don't do that the results are
interesting, they look intriguing and you would like to know more, but without
replication you can't admit it as evidence, (p. 141)

2. THE STUDY

One of the recognized characteristics of human ... [cultural history] is the tendency of
its successes ... to occur close together in relatively brief periods.
The quote is from Kroeber's Configurations of Culture Growth, published in
1944. Alfred L. Kroeber, one of the leading anthropologists of his time, was
apparently the first scholar who spent considerable research efforts on the
puzzling observation that the course of cultural evolution is discontinuous. The
Golden Ages of Renaissance, Baroque, Rococo, Romanticism, are the best-
known among Kroeber's "spurts of higher cultural creativity" in Western
civilization. He is fond of terms like "spurts" and "bursts" and phrases like
"pulses of production," and "geniuses turn up in clusters." Cultural growth was
no less discontinuous among the Greeks, the Romans, the Indians, Chinese,
etc. The phenomenon appears to be universal.
Kroeber was one of the first to use quantitative tools for historical research:
time series of counts of eminent contributors to human culture, being unaware
of similar procedures introduced earlier by Pitirim Sorokin (Sorokin, 1937-
494 Further Eysenckian interests

1941). His work, however, is almost exclusively descriptive, the question of why
geniuses pop up in clusters is barely addressed. He deliberately avoids
explanation because
we know extremely little, in any systematic and coherent way, about how [cultural
patterns] function and operate ... [The] why of their behaviour ... may ultimately lead
us back into psychology, or into the complex and obscure field in which psychological
and sociocultural factors are enlaced. But I deliberately refrain from any ultimate
explanation. (Kroeber, 1969, p. 19)
He did advance, though, a tentative explanation in terms of emulation:
Genius is fostered by emulation, and it is now envy, now admiration which enkindles
imagination ... In the beginning we are fired with the ambition to overtake those
whom we regard as leaders. [But] it is difficult to continue at the point of perfection
... Our zeal wanes with our hope. It ceases to follow what it cannot overtake, and
abandoning the old field as though pre-empted it seeks a new one. (p. 18)
Emulation implies role-models, the presence of top achievers as formal or
informal teachers for younger generations. Role model availability has been
extensively investigated by Dean Simonton (1988). Results from generational
time series analysis suggested the conclusion that Kroeber's theoretical
approach "is consistent with the facts" (p. 237). The link between creators
of successive generations (painters, poets, philosophers, etc.) manifested itself
by autoregressive coefficients.
There is a problem, however. A theory in terms of emulation and role model
availability predicts either no change of productivity at all (beyond chance
fluctuation) or a gradual and steady increase of productivity. Emulation
implies that cultural evolution is continuous and its pace gradual. Accelerating
tendencies might be stifled periodically by exhaustion whenever the "field
becomes pre-empted" and "the zeal wanes." Colin Martindale's pattern of
cultural evolution would probably fit in here, since it would predict wave-like
upward trends for reasons compatible with Kroeber's intuition (Martindale,
1990).
The main issue of interest, however, namely that pulses of creations might
occur without any role model instigation, is not accounted for. Kroeber is
aware of the problem, but he does not address it. Simonton seems to dismiss
the issue as he belittles Kroeber's configurations as mere "fluctuations" and
"probably due to white noise":
Most of the ups and downs are subject to random shocks. Even if many of these
haphazard inputs involve sociocultural conditions, a ... prominent portion could
concern the lucky influx of certain special personal characteristics ... By the luck of
the draw, some generations may shelter more persons with this rare confluence of
necessary traits injecting an era with more notable achievements than might be
anticipated according to the Zeitgeist alone. (Simonton, 1988, p. 237)
 Bursts of creativity 495

Simonton's view is based on his principle of "constant probability of success"

to which the present study is closely related. But our stance is almost opposed
to this principle. It is maintained here that the probability of cultural success is
inconstant, that historical variation is not white noise, and that there are clues
for reconsidering past approaches to the question: "Where do Kroeber's
configurations of culture growth come from?"
The present chapter provides an account of six observations. Taken together
they suggest a new direction in our search for antecedents.

2.1 The first observation

Bursts of cultural creativity are rare historical anomalies lasting for a couple of
decades.
This observation is based on two quantitative procedures, namely count of
creations and count of creators. The time series of our studies consists of
pentad units; occurrences of either creations or production scores are summed
over five years successively on the historical time axis. A pentad unit is
described as florescent for, say, paintings, if the count of paintings during the
respective five-year period is greater than expected. This is illustrated by
Figure 22.1. In his comprehensive reference work, the Belgian historian
Isabella Errera compiled the works of Western painters of all times and all
nations (TV = 40,700) extracted from about 1550 sources (exhibitions,
museums, sales, catalogues, etc.) (Errera, 1920/21). Figure 22.1 shows the

1100-

1000- Painting
900- -Works (Source: Errera)
• Exponential growth
800-

=1 =| 500-

300-

200-

100-

0-
(2)
1400 1450 1500 1550 1600 1650 1700 1750 1800
A. D.
Figure 22.1. The time series of Western paintings for A.D. 1400-1800, based on a comprehensive
index (Errera), reveals two bursts of productivity (see vertical dotted lines).
496 Further EysencJdan interests

Figure 22.2. Counting productive years of creators—without considering actual creations—serves

as a substitute for counting creations. The method is exemplified by five fictitious creators (for
details see text).

distribution of the Errera database paintings from A.D. 1400 to 1800, by

pentad resolution. Two abrupt pulses of creativity are revealed which clearly
exceed the line of exponential growth.
The curve of Figure 22.1 has been obtained by counting productions. The
method of counting production scores is illustrated by the lower part of Figure
22.2. On the time axis, birth years of six fictional producers are marked by up-
arrows, death years by down-arrows. Following the results of Simonton (1982)
a person's flourish period is assumed to occur on the average hi his mid-life, a
section of 15 years, from age 35 to 49, was therefore taken as the most probable
period of his productivity. For example, the first person, bora in 1590,
 Bursts of creativity 497
550

500- Science
450- - Production scores (Source: Asimov)
400- Exponential growth

350-

250-

200-

150-

100-

50-

1400 1450 1500 1550 1600 1650 1700 1750 1800

A. D.
Figure 223. The time series of production scores of scientists in Western civilization reveals two
bursts of creativity comparable to those for painting of Figure 22.1 (see vertical dotted lines).

contributes one productivity score to each year from 1625 to 1639 (see the gray
bar on the time axes). Productivity scores were summed across persons for
each pentad. The distribution in the upper part of Figure 22.2 shows a time
series of productivity for the six fictional creators.
The summing of production scores was applied, to each of the databases
considered here, for example, to the total of scientists excerpted from Asimov's
biographical dictionary (Asimov, 1982). Figure 22.3 shows the distribution of
scientifically productive pentads from A.D. 1400 to 1800. The line of
exponential growth is here surpassed again by two conspicuous peaks.
Do production scores (estimates) indicate actual production rates? Since
production rates in the first half of this period (i.e., 1400-1600) have too many
missing cases, production rates from the second half only (i.e., 1600-1800)
were compared with production scores (see Figure 22.4, p. 499). The time
series of productions (the solid line) was obtained by collecting entries from 26
reference works (see the Master Index). The dashed curve shows production
scores obtained from five sources: Asimov (1982), Eisler (1912), Gillispie
(1970-1980), Williams (1982), ZiegenfuB (1949-1950). It can be seen that the
two series covary considerably.
A slight difference shows up as one should expect: The peak of creativity is
more marked when based on actual works than on production scores. The
latter appear to be diluted by individual differences. Not all creators enjoy their
most productive life time from age 35 to 49. Nevertheless, the correspondence
between the two types of counts is satisfactory; production scores, based on life
data of creators, appear to be reliable and valid. This conclusion is also backed
498 Further Eysenckian interests

by comparing production scores of Figures 22.3 (based on one source) and 22.4
(based on five sources), which are highly correlated (r = .90, residuals of
exponential growth).
Cross-correlations between paired detrended time series drawn from
biographical sources of comparable size are generally considerable.

2.2 The second observation

Bursts of creativity co-occur jointly in different cultural fields.
The preceding curves revealed peaks of exceptional florescence for painting
and for science. Poetry curves display corresponding discontinuities, as will be
shown later. The co-occurrence of creativity bursts in different cultural fields
suggests a common cause. The question arises: Do they occur at exactly the
same time? The answer is "No." Pertinent evidence is based on the third
observation (see below).

2.3 The third observation

Bursts of creativity in different fields occur in regular succession.
This becomes evident by comparing seventeenth century works of painting
(Errera, 1920-1922), poetry (H. A. Frenzel & E. Frenzel, 1980), and science
(Master Index) (see Figure 22.5). Even though peaks do occur in the three
fields, they are not precisely in phase. They appear to succeed one another by a
10-15-year phase difference.
First comes painting, then poetry, then science. The same succession turns
up with the sixteenth century peaks. During this period, however, the phase
distance between peaks seems to be twice as large.

2.4 The fourth observation

Bursts of creativity occur simultaneously in independent cultures.
This observation is puzzling and challenging. Among the non-European
civilizations of our study (i.e., Chinese, Japanese, and Ottoman) the Chinese
provide the most reliable and most extensive biographical records. In Figure
22.6, production scores of Chinese scientists and thinkers (dashed line, source:
Chen, 1976, A.D. 1600-1800) are compared with those of Western scientists
and thinkers (solid line) (sources: as for Figure 22.4).
Note that there is a peak of creativity in China, and that this peak occurred
at the same time as the European peak. The only difference is that the Chinese
curve does not show an abrupt rise in the eighteenth century compared to the
European curve.
 Bursts of creativity 499
200

180- Science
- Production scores (Source: 5 bfogr. dictionaries)
160- -Works (Source: Master Index, 26 titles)

140-

100-

60-

40-

20-

0
16001625165016751700172517501775 1800
A. D.
Figure 22.4. Changes of production scores of scientists reflect changes of actual productions
(discoveries, inventions) to a large extent.

Synchronous discontinuities are also apparent in time series of European

and Chinese painting (see Figure 22.7).
How to explain the dent on top of the Chinese peak of painting? This may be
tentatively explained by an accident in Chinese history: In A.D. 1645, the
Mongolian Manchu conquered China's mainland, overthrew the Ming dynasty,

200
Paintings (Errera, EUR)
180- Poems (Frenzel, GER))
Scientific discoveries (Master Index, EUR)
160-
140 Painting
/X Poetry
12<
100
'i

16001625165016751700172517501775 1800
A. D.
Figure 22.5. The seventeenth-century discontinuities in painting, poetry, and science are not
exactly synchronous: midcentury peaks succeed one another by 10-15 years.
500 Further Eysenckian interests

200
180 Science & Philosophy
- European (Source*: 6 btogriplfcal, 1 works)
160 -Chinese (Source: Chen)

140
£^J20
f f 100J
2* SO-
IL
60
40-
20-
0
160016251650167517001725175017751800
A. D.
Figure 22.6. Time series based on Chinese and European science data reveal bursts of creativity in
both cultures and at exactly the same time.

and tried to demolish its tradition. Even its books were burned. No dent was
observed on top of the Chinese science/philosophy peak of Figure 22.5. This is
probably because, as was shown earlier, creativity in science boomed 25-30
years later. The violent acts of the new rulers had ceased by that time.

200
180- Painting
-European (Source:6btogr.dfcttonerte*)
160- -Chinese (Source: 5 btogr. dfctioneries)
140-
.f^ 120-
II 100 s
r °i 60-
40-
20-
0
160016251650167517001725175017751800
A.D.

Figure 22.7. Concurrent discontinuities in the time series of Chinese and European painting is
another challenging observation.
 Bursts of creativity 501

2.5 The fifth observation

Bursts of creativity are independent of total numbers of creators.

This observation may be explained by Simonton's principle of "constant

probability of success." According to this researcher, who adopted the model
of Campbell (1960), "the odds of making a contribution are probabilistic
consequences of total output." Therefore, "both major and minor works tend
to covary ..." (Simonton, 1988, p. 409). In other words: the greater the number
of ideas in a person's head, the greater is the probability of exceptionally good
ones among them. Likewise, the greater the number of producers of ideas in a
society, the greater is the number of geniuses among them.
For science, a time series was obtained by assembling all entries in Gillispie's
monumental 16 volumes of biographical articles providing a rough estimate of
the science expert population (dashed line in Figure 22.8, source: Gillispie,
1970-1980). Another time series was based on Asimov's (1982) and Williams'
(1982) biographical dictionaries of eminent scientists (solid line in Figure
22.8).
The solid curve shows that the seventeenth century rise of eminent scientists
is hardly related to the fluctuation in size of the total expert population. In
addition, an explanation in terms of mere quantity would not account for the
synchronous florescence of science in independent cultures. Could a
civilization ever suddenly multiply, by a factor of 2 or more, and for a couple

400
(European)
350-
Productive years:
High eminent. Source: Asimov, Williams
300- Less eminent. Source: Gillispie

250-

fi
1*
Q_
200-

150-
100-
50-
0
16001625165016751700172517501775 1800
A. D.
Figure 22.8. A spurt of productivity is apparent for the minority of highly creative scientists, the
productivity of the majority of less eminent scientists is far less discontinuous.
502 Further Eysenckian interests

of decades only, the numbers of experts in all branches of cultural production?

Moreover, could this happen in synchrony with some unknown civilization
10,000 miles away?
Thus, discontinuous culture growth cannot be explained by varying
population size. Four other attempts at explaining cultural florescence,
without positive results were done by Narrol et al., (1971). He found that
neither the wealth of civilizations nor then" geographical expansion, nor
political rulership, nor challenge by warfare explained cultural florescence. The
only significant factor was political fragmentation; this variable proved its
effect in an independent study by Simonton (1975). This author found
additional factors fostering generational fluctuations of creativity, namely
political and imperial instability. While this evidence is appreciable, it does not
account for synchronous fluctuations in unrelated civilizations.

Suggested factors explaining cultural creativity

Explanation required for (1) (2)

Discon- Synchronous
tinuities independent
discon-
tinuities

B Biological - Genetic variation

.£ Psychological -Emulation
^ (Kroeber)

Demographical - Role-modej
availability
Political Fragmentation
(Nan*. Simonton)

I
"o
Instability
(Simonton)
o Warfare
CO (Nan*)
Expansion
(Nan*)
Economical Wealth
(Nan*)

Supra-Societal
(Chizhevsky-Ertel)

Figure 22.9. Individual and societal factors have been proposed to explain intrasocietal cultural
fluctuations, with partial empirical confirmation. Synchronous cross-societal fluctuations of cultural
productivity, however, can only be explained by supra-societal ("macroecological") factors.
 Bursts of creativity 503

Figure 22.9 surveys hypothetical determinants as have been proposed to

date.
Column 1 represents hypothetical factors explaining within-cultural varia-
tion, column 2 shows factors explaining concurrent cultural variation. A rough
division between individual and societal factors is made (see the left margin).
Under column 1, for example, Kroeber's emulation is coded as an individual
(psychological) factor; the plus mark indicates that it has been empirically
confirmed.
Beneath emulation are listed societal factors tested by Simonton and Narrol
et al., three of which contribute to creative fluctuations. The problem is that
none of these factors, even those with plus signs in the first column, could ever
bear on the second column, that is, the simultaneous ups and downs of cultural
evolution in noninteracting civilizations (see the minus signs on the right).
In view of this observation, it seems unavoidable to postulate some supra-
societal factor operating on different societies at the same time (see the bottom
line). Fortunately, the bottom line need not be left entirely blank. There is
another observation, one that suggests an unusual hypothesis, but, without
competing alternatives, this observation warrants consideration.

2.6 The sixth observation

Bursts of creativity correlate with solar activity.
This observation is related to heliobiology, a pioneering field of research in
biology and medicine, originated by Chizhevsky (1924). The present status of
this interdisciplinary field of research cannot be summarized here. Suffice it to
say that the pertinent literature is extensive, though not widely known (Sigel,
1975; Dubrov, 1978; Smith & Best, 1989; Tomassen, de Graaf, Knoop,
Hengevold, 1990). The author has dealt with this hypothesis previously in Ertel
(1988, 1991, 1994, 1996).
The dark spikes in Figure 22.10 represent counts of sunspots, visible
symptoms of solar magnetic variation. The solid curve is the running mean of
solar activity. As can be seen, solar activity is cyclic but not entirely regular. The
most conspicuous irregularity is the sudden dwindling of the sunspot amplitude
starting at about A.D. 1625. Solar activity even ceases almost entirely from
A.D. 1645 to 1715. Astronomers refer to this phenomenon as an "anomaly" of
an otherwise "well-behaved" sun whose rest period at that time is called the
Maunder Minimum. The Astronomer Maunder discovered it in 1898, but his
peers acknowledged it only 80 years later when the evidence had become
overwhelming. Astronomers have not yet explained the phenomenon [see
Eddy (1977, 1983)]. A less conspicuous dip in the sun's average amplitude
occurred around 1790-1820, the Eddy Minimum.
504 Further Eysenckian interests

Figure 22.10. A striking decline of solar activity (sunspots) in the seventeenth century (Maunder
Minimum) and a slight dip of its amplitude at the turn of the eighteenth to nineteenth century
(solid line) is reflected by concurrent chances of radiocarbon production in the earth's atmosphere.
The present-day radiocarbon time series go far back into the past.

Sunspots were not observed and recorded before the telescope was invented
(A.D. 1610). Fortunately, modern geochemistry provides radiocarbon time
series (see the dashed curve in Figure 22.10) replacing, even though with less
precision, the method of sunspot counts. Radiocarbon measures are generally
considered to be valid indicators of average solar activity levels. A millennia-
long record is available today.
Another look at seventeenth century science/philosophy and painting shown
earlier (see Figures 22.1 and 22.3) is given by Figure 22.11 with the Maunder
Minimum period inserted (see the dashed lines). Note that the burst of
creativity in science occurred after the onset of the solar excursion. If solar
activity is causally involved, dependent phenomena should succeed solar-
terrestrial variation as it does here. The burst of creativity in painting occurred
at a time when solar activity went down to its exceptionally low amplitude. A
causal relationship is conceivable here, too, painting may "react" sooner than
science—for whatever reasons.
Results of a third and final test is presented in Figure 22.12 covering the full
observational period of 400 years. The curve above represents 14C values, an
operational measure of mean solar behavior. The pertinent scale on the right
has been inverted so that a decrease of solar activity is represented by an
upward change, an increase by a downward change. The higher the curve, the
less active the sun.
 Bursts of creativity 505

A.D.
1600 1625 1650 1675 1700 1725 1750 1775 1800

1600 1625 1650 1675 1700 1725 1750 1775 1800

A. D.

Figure 22.11. Changes of cultural productivity in China and Europe (data from Figures 22.6 and
22.7, painting and science) occurred at the time when the Maunder Minimum started or soon
thereafter.

As can be seen, radiocarbon dating reveals another solar excursion in the

fourteenth and fifteenth century, the so-called Sporer Minimum. Sporer was
an independent discoverer of the anomaly in solar behavior without receiving
recognition among his peers. Thus, across A.D. 1400-1800, solar activity
showed two anomalies.
Are solar discontinuities related to discontinuities in human cultural history?
Figure 22.12 provides production scores for writers, dashed for European
writers (source: Bertelsmann's Lexicodisc, 1992), solid and bold for Chinese
writers, of all writing professions (source: Chen, 1976). It appears that science
and painting covaries with solar activity in the same way that literature does.
Only one difference is noticed: The European (dashed) curve below, after
506 Further Eysenckian interests

—' d4 • * Chinese • —— European

(*) SpOrer-Minim. Maunder-M.

200

1400 1450 1500 1550 1600 1650 1700 1750 1800

A.D.

Figure 22.12. Another solar anomaly (Sporer Minimum) preceded the Maunder Minimum (see
dotted line above). Chinese and European productivity of writers, almost concurrent (lines below),
may be correlated with solar activity. Two oddities ensue which could be explained by a nonlinear
relationship between solar-environmental factors and cultural productivity implying an optimum
effect with reduced, but nonzero solar "stimulation."

having descended in the post-Sporer decades, resumes its upwards trend

earlier than the Chinese curve (solid). This is true for literature, not for
painting and science. Historical case studies might clarify this minor departure.
Two oddities occurring in both civilizations arouse additional attention.
First, even though cultural productivity rises in both, the Sporer and Maunder
Minima, the onsets of their decline differ. In the Sporer Minimum productivity
rises up to the Minimum's close, while in the Maunder Minimum the
productivity declines about 30 years prior to the Minimum's close. This
difference may be due to the fact that the dearth of solar activity in the
Maunder period was more extreme than in the Sporer period. A moderately
low level of solar activity seems to be an optimal condition for cultural
florescence (see dashed horizontal line above). Zero solar activity, which
apparently prevailed in the second half of the Maunder period only, may
indicate conditions detrimental to cultural florescence. This speculation is
backed by evidence obtained for solar effects by short-term 11-year oscillations
(see Ertel, 1988).
The second oddity is this: cultural productivity decreases in post-Sporer
decades while it increases in post-Maunder decades. The difference is possibly
related to the above observation: the close of the Sporer Minimum might have
brought optimal helioterrestrial conditions to an end; subsequent productivity
 Bursts of creativity 507

would therefore decrease, while the close of the Maunder Minimum should
have terminated detrimental conditions, therefore an increase of productivity
would ensue.

3. CONCLUSIONS

Further investigations into the above speculations are needed. Nevertheless,

the fact that the oddities occurred independently in both cultures suggests that
an explanation in terms of macro-ecology might apply even here, at least in
principle. At this early stage of heliobiological research, any hypothesis
involving solar system conditions is speculative. What appears to be an oddity
today may be resolved in the future by an improved understanding of solar-
terrestrial relations. Hopefully, the present ignorance regarding the nature of
helio-dependent stimulations and of their impact on creative responses will
eventually diminish, and an underlying neuropsychological mechanism might
be discovered. Doubts and open questions are generally prolific, and these
usually stimulate research. It remains true that "it is clearly too early to say
whether Kroeber's question has been fully answered" (Simonton, 1981, p. 629).
I, too, refrain here, as did Simonton and Kroeber himself, from premature
conclusions expecting surprising advances from extended interdisciplinary
endeavors.

REFERENCES

Asimov, I. (1982).Asimov's biographical encyclopedia of science and technology (2nd rev.

ed.). Garden City, NY: Doubleday.
Benski, C, Coudron, D., Galifret, Y. et al. (1996). The "Mars Effect."A French test of
over 1,000 sports champions. Amherst, NY: Prometheus.
Campbell, D. (1960). Blind variation and selective retention in creative thought as in
other knowledge processes. Psychological Review, 67, 380-400.
Chen, C. K. H. (1976). A biographical and bibliographical dictionary of Chinese authors (3
vols.). Hanover, NH: The Oriental Society.
Chizhevsky, A. L. (1924). Physical factors of the historical process (in Russian). Kaluga:
Gostinopolygrafic.
Dubrov, A. P. (1978). The geomagnetic field and life: Geomagnetobiology. New York:
Plenum.
Eddy, I. A. (1977). Climate and the changing sun. Climatic change, 1, 173-190.
Eddy, I. A. (1983). The Maunder Minimum. A reappraisal. Solar Physics, 89, 195-207.
Eisler, R. (1912). Philosophen-Lexikon. Berlin: Mittler.
Errera, I. (1920-21). Repertoire des peintures datees (2 vols.). Brussels: Libraire
Nationale d'Art et d'Histoire.
508 Further Eysenckian interests

End, S. (1986) Interview mit Hans-Jurgen Eysenck fiber die Grenzgebietsforschung.

Zeitschrift fur Parapsychologie und Grenzgebiete der Psychologic, 28, 136-155.
Ertel, S. (1988). Scientific discoveries: Testing for concomitance with extraterrestrial
events. Paper read at the 7th Annual Meeting of the Society for Scientific
Exploration, 2-4 June 1988, Ithaca, NY. (Unpublished).
Ertel, S. (1991). Patterns of scientific evolution: Short-term cycles and secular waves. In
Best, H., Mochmann, E., & Thaller, M. Computers in the humanities and social
sciences. Achievements of the 1980s. Prospects for the 1990s. Munich, Germany: K. G.
Saw.
Ertel, S. (1994). Influenza pandemics and sunspots—easing the controversy. Natur-
wissenschaften, 82, 308-310.
Ertel, S. (1996). Space weather and revolutions. Chizevsky's heliobiological claim
scrutinized. Studia Psychologica, 38, 3-22.
Ertel, S. & Irving, K. (1996). The tenacious Mars effect. London: Urania.
Eysenck, H. J. (1975). Planets, stars, and personality. New Behaviour, 29 May, 246-249.
Eysenck, H. J. (1978). On Jerome's "Astrology Disproved." Phenomena, 2(2) (March-
April), 9-13.
Eysenck, H. J. & Nias, D. (1982). Astrology, science or superstition? London: Maurice
Temple Smith.
Frenzel, H. A. & Frenzel, E. (1980). Daten deutscher Dichtung: Chronologischer Abrifi
der deutschen Literaturgeschichte (2 vols.). Munich, Germany: Deutscher Taschen-
buchverlag.
Gillispie, C. C. (ed.) (1970-80). Dictionary of scientific biography. New York: Scribner.
Kroeber, A. L. (1944). Configurations of culture growth. Berkeley, CA: University of
California Press.
Martindale, C. (1990). The clockwork muse. The predictability of artistic change. New
York: Basic Books.
Naroll, R., Benjamin, E. C., Fohl, F. K. et al. (1971) Creativity: A cross-historical pilot
survey. Journal of Cross-Cultural Psychology, 2, 181-188.
Sigel, F. (1975). Schuld ist die Sonne. Frankfurt: Hani Deutsch Thun.
Simonton, D. K. (1975). Sociocultural context of individual creativity: A transhistorical
time-series analysis. Journal of Personality and Social Psychology, 37, 1119-1133.
Simonton, D. K. (1981). Creativity hi Western civilization: Intrinsic and extrinsic causes.
American Psychologist, 36, 628-629.
Simonton, D. K. (1983). Creation productivity and age: A mathematical model based on
a two-step cognitive process. Developmental Review, 3, 77-111.
Simonton, D. K. (1988). Creativity, leadership, and chance. In R. J. Sternberg (Ed.),
The nature of creativity. Cambridge: Cambridge University Press.
Simonton, D. K. (1988). Galtonian genius, Kroeberian configurations, and emulation: A
generational time series analysis of Chinese civilization. Journal of Personality and
Social Psychology, 55, 230-238.
Smith, C. W. & Best, S. (1989). Electromagnetic man. Health and hazard in the electrical
environment. London: Dent.
Sorokin, P. (1937-41). Social and cultural dynamics (4 vols.). New York: American
Book Co.
Tomassen, G. J. M., de Graaf, W., Knoop, A. A., & Hengevold, R. (1990). Gee-cosmic
relations; the earth and its macro-environment Proceedings of the First International
 Bursts of creativity 509

Congress on Geo-cosmic Relations. Foundation for Study and Research of

Environmental Factors, Amsterdam, 19-22 April 1989. Wageningen, The Nether-
lands: Pudoc.
Williams, T. I. (Ed.) (1982). A biographical dictionary of scientists. New York: Wiley.
ZiegenfuB, W. (Ed.) (1949-50). Philosophen-Lexikon. Handworterbuch der Philosophic
nach Personen. Berlin: de Gruyter.

MASTER INDEX

The Master Index of scientific discoveries and technological inventions is based

on the following 26 sources:
Aschoff, L., Diepgen, P., Goerke, H. et al. (1960). Kurze Ubersichtstabelle zur Geschichte
der Medizin (7th edn.). Berlin: Springer.
Brockhaus (1971). Brockhaus der Naturwissenschaften und der Technik (4th edn.).
Wiesbaden, Germany: Brockhaus.
Canby, E. T. (1963). Geschichte der Elektrizitat. (Vol. 7.) Lausanne: Editions Rencontre.
Collier's encyclopedia (1964). New York: Collier. (Index entry: Inventors and invention).
Daintieth, J. et al. (1981). A biographical encyclopedia of scientists (Vol. 2). New York:
Facts on File.
Grun, B. (1982). The time tables of history. New York: Simon & Schuster.
Hellemans, A. & Buisch, B. (1988). The time tables of science. New York: Touchstone.
Herder (1979). Herder-Lexikon Naturwissenschaftler. Freiburg, Germany: Herder.
Hermann, A. (1987). Lexikon Geschichte der Physik. Cologne, Germany: Aulis.
Lane, H.-U. (1980). The world almanac book of who. New York: World Almanac.
Liider, E. (Ed.) (1963). Kleine Enzyklopadie Technik (3rd edn.). Leipzig: Bibliogra-
phisches Institut.
Martin, I. (1984). Le livre mondial des inventions. Paris: Bureau d'Enregistrement des
Inventions.
Mumford, L. (1947). Technics and civilization (6th edn.). London: Routledge.
Parkinson, C. L. (1985). Breakthroughs. A chronology of great achievements in science and
mathematics 1200-1930. London: Mansell.
Peters, A. (1980). Synchronoptische Weltgeschichte. Munich, Germany: Universum.
Reichen, C.-A. (1963). Geschichte der Astronomic. (Vol. 5.) Lausanne: Editions
Rencontre.
Reichen, C.-A. (1963). Geschichte der Chemie. (Vol. 10.) Lausanne: Editions
Rencontre.
Reichen, C.-A. (1963). Geschichte der Physik. (Vol. 8.) Lausanne: Editions Rencontre.
Rothschuh, K. E. (1952). Entwicklungsgeschichte physiologischer Probleme. Munich,
Germany: Urban & Schwarzenberg.
Schneiderbauer, H. (1961). Die groflen Erfindungen der Welt. Munich, Germany:
Siidwestverlag.
Starobinski, J. (1963). Geschichte der Medizin. Lausanne: Editions Rencontre.
Stein, W. (1976). Kulturfahrplan. Munich, Germany: Herbig.
Sworykin, A. A. (1964). Geschichte der Technik. Leipzig, Germany: Fachbuchverlag.
510 Further Eysenckian interests

Valentin, M. (1950). Geschichte der Pharmazie und Chemie. Stuttgart, Germany:

Wissenschaftliche Verlagsanstalt.
Walden, P. (1952). Chronologische Ubersichtstabellen zur Geschichte der Chemie. Berlin:
Springer.
Wetterau, B. (1990). The New York public library book of chronologies. New York:
Stonesong.

Acknowledgments
Most of the databases were compiled by student assistants whose contribution
to the study is acknowledged: Claudia Brand, Michael Epple, Barbara Fricke,
Beate Glaser, Stefan Gunkel, Kerstin Karg, Claudia Kasten, Christian
Kordowski, Anja Krause, Susan Lathe, Christof Linnemann, Matthias
Romppel, Peter Ruhlender, Wiebke Sander, Tatjana Schnell, and Britta
Wahle. David Valentiner gave helpful suggestions for improving the chapter.
Chapter 23
Graphology, astrology, and parapsychology

G. A. Dean, D. K B. Nias, and C. C. French

The weight of the proofs must be suited to the oddness of the facts. Laplace

1. INTRODUCTION

This chapter looks at Eysenck's involvement with graphology, astrology, and

parapsychology, which led to a total of nearly 50 publications (see Table 23.1).
In each case the result was a confrontation between a learned man of science
and areas where convincing evidence is hard to come by. He was, as Dwyer
(1986) an astrological editor put it, "a worldwide known critic of insufficiently
demonstrated assertions, as well in classical psychology as in borderline
science." The present chapter is an update of an earlier survey by Nias and
Dean (1986) that appeared in Hans Eysenck: Consensus and Controversy, the
second volume of the Palmer International Master-Minds' Challenged Psychol-
ogy Series (Modgil & Modgil, 1986). In his overview of that volume, Gibson
(1986) noted that the Nias and Dean survey was "a serious effort to discuss the
surprising fact of Eysenck's involvement with the whole area, an involvement
that has surprised so many of those who take his work very seriously" (p. 5).
Gibson's (1981, pp. 193-215) own lengthy survey of Eysenck's involvement
with astrology and parapsychology provides historical background and details
of the critical studies up to that time.

1.1 What is so attractive about borderline science?

Each of our three areas (graphology, astrology, parapsychology) is contro-
versial but each has a solid core of testable ideas. It was this important quality
of testability, plus apparently positive results, that first attracted Eysenck's
attention. As Eysenck (1990) says in his autobiography, "the main attraction of
a field such as hypnosis (or parapsychology, or astrology), is the promise it
contains of entirely new and extremely important knowledge that might be
gained by a study of the (alleged) phenomena in question" (p. 233).
512 Further Eysenctdan interests

Table 23.1. Frequency of Eysenck's publications in borderline science by decade

Decade Graphology Astrology Parapsychology Total

1945-54 2 0 0 2
1955-64 2 0 0 2
1965-74 0 0 3 3
1975-84 0 17 5 22
1985-94 1 10 4 15
1995- 0 2 1 3
Total 5 29 13 47

The figures show how the focus, initially on graphology, changed after two decades to include
parapsychology and then astrology. There is a pronounced peak in 1982-1988 following
publication of Astrology: Science or Superstition1}, which led to numerous articles on methodology
and on the Gauquelin results, most of them published in journals of astrological research.
Entries are the total of journal articles, books, anthology chapters, and book reviews, all listed in
the references. Not included are a few works in foreign languages. The books and anthologies are:
Sense and Nonsense in Psychology, 1957; Handbook of Abnormal Psychology, 1960; Encyclopedia
Britannica, 1974; Astrology: Science or Superstition? 1982; Explaining the Unexplained, 1982;
Research in Parapsychology, 1983; Know Your Own Psi-Q, 1984; Social Science Encyclopedia, 1985;
Hans Eysenck: Consensus and Controversy, 1986; Explaining the Unexplained, 2nd edition, 1993.

Eysenck (1986) emphasizes that he came to his positive conclusions (on the
Gauquelin findings, ESP, and psychokinesis) only after carefully assessing the
evidence and failing to find sources of error. He comments
it is said that one should not waste time on topics which are obviously absurd,... I do
not believe myself that a priori judgments of this kind are admissible in science;
scientists have been wrong too many times in making explicit statements of this land
to be considered infallible. In any case, the time that is wasted is mine, and to waste it
by reading the literature on astrology and parapsychology is probably better spent
than in watching pornographic films, or becoming a football hooligan! (pp. 382-383)
Eysenck's positive conclusions arose even though the claims seemed at first
unlikely to be true. The result was appreciable cognitive dissonance:
I do not enjoy having to defend empirical findings which go counter to my own
instinctive beliefs. I would much rather be in a position to disprove all
parapsychological and astrological claims: life would be so much easier if we could
cosily go to sleep in the shadow of orthodox science!
But the facts decided the issue:
I certainly did not come to positive conclusions in these matters simply in order to
annoy orthodox scientists ... Perhaps strong innate feelings for the underdog have
something to do with it; I believe that these fields have been decried by orthodox
scientists without specialist knowledge of what has been done in them, and this I
consider to be insupportable (p. 384)
 Graphology, astrology, and parapsychology 513

Furthermore, as Eysenck and Nias (1982, p. viii) point out, one of the few
reliable and valid generalizations in social psychology is the "principle of
certainty" first stated in 1935 by R. H. Thouless of Straight and Crooked
Thinking fame. Namely that when there is evidence both for and against a
belief, most people do not show low levels of conviction, which logically they
should, but instead they show high levels of conviction either for or against,
which logically is indefensible. The principle applies to all kinds of social
attitudes including religious beliefs, and helps to explain the controversy
attached to our three areas. Which makes the arrival of Eysenck the man of
science all the more interesting.

1.2 The forgotten variable: effect size

Utts (1991,1996), among others, has pointed out that significance level is not a
measure of effect size, nor of whether the sample size is sufficient to detect it.
As noted by Nias and Dean (1986) "the long runs favoured by [psi] researchers
can inflate trivial effects to impressive significance. For example, one's
astonishment at a test of 40,000 coin tosses that produced evidence for psi at
the 0.001 level might well evaporate on our learning that for every 100 tosses it
required averaging 50.8 heads instead of the 50 expected by chance" (p. 374).
One common way (adopted here) of expressing effect size is as a correlation;
for example, Pearson r for paired observations, phi for 2x2 tables, and kappa
for hits and misses (Cicchetti, 1987), where kappa = (observed hits - expected
hits) / (N — expected hits).
Until very recently the traditional emphasis on significance level has applied
in graphology, astrology, and parapsychology. But the crucial point is that in all
three areas the effect sizes as determined by meta-analysis are very small,
around 0.1 or less (see Table 23.2). Eysenck (1978, 1994) has in no uncertain
terms noted how meta-analysis becomes mega-silliness if no allowance is made
for study quality. Later we show that such concerns do not affect the general
picture shown in Table 23.2.
Glymour (1987) argues that no sensible person will opt for a paranormal
explanation of tiny effect sizes. He points out that these could have many
ordinary causes, most of them impractical to control even if known. So a
sensible person will say that tiny effect sizes are more probably due to some
combination of ordinary causes. In effect this is a wielding of Occam's Razor,
where (as restated by Bertrand Russell) "entities are not to be multiplied
without necessity." But Eysenck has often pointed out that the history of
science is full of surprises. He would say that an effect size, however small,
deserves an explanation if it is significant and repeatable. We agree. In fact
everything hangs on repeatability, for without repeatability we cannot vary the
associated variables to find out what might be happening.
514 Further Eysencldan interests

Table 23.2. Effect sizes in graphology, astrology, and parapsychology, with some other effect sizes
for comparison

Graphologists' judgments 0.12 (0.11) N= 107

Nongraphologists' judgments 0.12 (0.08) N = 34
Agreement between graphologists 0.42 (0.14) N = 15

Astrologers' judgments 0.05 (0.12) N = 40

Agreement between astrologers 0.10 (0.06) N = 25
Gauquelin planetary effects 0.04 (0.01) N = 10
Sun-sign prior knowledge 0.09 (0.03) N= 19

ESP vs. random number generators 0.0003 AT = 597

Forced-choice precognition 0.020 N = 309
Auto-ganzfeld studies (see Fig. 23.3) 0.12 N = 11
Psychokinetic effects on dice 0.012 TV = 148
ESP hit rate vs. extraversion 0.14 N = 60

IQ scores between identical twins 0.72 N = 119

EPI scores vs. self ratings, peer ratings 0.56 N= 13
Law school ability test vs. first-year law grade 0.34 N = 726
Physiognomy vs. IQ, personality test, peer ratings 0.15 N=17
Unstructured interviews vs. work performance 0.11 N=19
Age (adults only) vs. work performance -0.01 N = 425
Palmistry vs. personality test, self-ratings -0.05 N =9
Each effect size is expressed as a correlation uncorrected for attenuation, and is generally based on
the meta-analysis of available studies. N is the number of studies. Figures in parentheses are
standard deviations. Rosenthal and Rubin (1982) point out that an effect size of r is equivalent to
getting 50(1+r)% hits vs. 50% expected. Sources for graphology Dean (1992), for astrology Dean
et al. (1996), for parapsychology Utts (1991), and for the rest Dean (1992).

2. GRAPHOLOGY
2.1 Introduction
The first borderline area to attract Eysenck's interest was graphology, the
divining of personality (and almost everything else) from handwriting. Eysenck
(1957) describes how his involvement began in 1934 when, among a gathering
of fellow students in Dijon, he casually mentioned he was a graphologist (which
he was not, but he wanted to keep the attentions of a certain young lady):
The response was literally overwhelming. Everybody pulled letters and other
documents out of their pockets, demanding that I should tell the character of the
people concerned ... About 95 per cent of the "clients" reported themselves amazed
at the uncanny accuracy of my characterizations. I think any scientist who would have
come in to challenge the claims of graphology at that time would have had a very
hard time indeed, (p. 223)
 Graphology, astrology, and parapsychology 515

The uncanny accuracy was, of course, due to his shrewd use of Barnum
statements, statements so general or so vague that they will fit anybody; for
example, "you have a good sense of humor," which at least 98% of the
population will admit to. Later, in 1945, Eysenck found graphology worth a
second look.

2.2 Eysenck's early studies on graphology

If the claims of graphologists ... have any justification, it is clear that the scientific
study of this discipline must be of outstanding importance to abnormal as well as to
experimental psychology. The possibility of obtaining ... [historical diagnoses and
personality analyses free from falsification]... of following the progress of the cure, or
of the effects of the experiment, in the image of the patient's or subject's
handwriting—these are advantages enjoyed by no other method of personality
study. (Eysenck 1945)
In those days few empirical studies existed. The only critical review was by
Allport and Vernon (1933), who noted that the average graphological reading
is especially difficult to validate owing to its unscientific and unstandardized
terminology;... Verbal self-contradictions appear frequently... and the terms employ-
ed often seem to obscure rather than reveal the personality. In some of our
investigations one-half to two-thirds of the points made by professional graphologists
had to be discarded as incapable of objective confirmation or denial, (p. 210)
Eysenck's (1945) own research was prefaced by a review of the 30 studies
that were "the most important 10%" of those which had appeared since 1933,
of which only ten reported effect sizes. Despite his enthusiasm for the potential
advantages of graphology, his doubts concerning the actual evidence were
made clear:
The boundlessness of the graphologist's faith, the enormity of his claims, must make
the cautious scientist hesitate ... much of the evidence as reported in the various
papers quoted cannot be regarded as more than suggestive. Too frequently the
controls have not been sufficiently stringent, the number of handwritings used too
small to give results free from serious sampling errors, and the criteria for validation
themselves too much lacking in both reliability and validity to make comparisons
fruitful, (pp. 70, 72)
In his own study, Eysenck (1945) had an experienced graphologist answer a
28-item personality test for 50 neurotic patients based on their handwriting,
predict their intelligence on a 5-point scale, and match case histories to
handwriting. He also had nongraphologists match case histories to the
graphology readings and handwritings. The results showed modest support
for graphology, but as Wolfson (1951, pp. 423-424) later noted, the success
could have been inflated by knowing the subjects were patients. Three years
later, Eysenck (1948) had another experienced graphologist predict
516 Further Eysencldan interests

neurotirism on a 5-point scale from the handwriting of 176 neurotics and non-
neurotics. Their psychiatric diagnosis did not correlate with the graphological
predictions (r = —0.02), whereas it did with a psychometric test battery (r =
0.73). He concluded:
The present study demonstrates that even with an imperfect criterion, short objective
tests show much higher validity coefficients than does graphological analysis; this
would seem to invalidate the claims of the graphologists. It would also seem to argue
against the idiographic view that "global" appraisals of personality are superior to
nomothetic or "atomistic" tests; handwriting analysis is a typical idiographic or global
procedure, while the series of tests used in this study might be considered
representative of the nomothetic approach, (p. 96)
Eysenck was careful to stress that this was merely one test with one
graphologist. But hi the end, for graphologists in general, it would be further
research that would decide the matter. He then put the ball hi their court:
the possibility that others might have been more successful cannot be ruled out,
although in our view it is not a very likely one ... the handwritings [are] available to
any graphologist who believes that the conclusion ... is too severe. By thus submitting
himself to the experimental test the critical graphologist would find an obvious way to
disprove the conclusion indicated above.

2.3 Eysenck's later studies on graphology

After this, Eysenck returned to graphology only sporadically. In 1960 he
allowed a review of graphology (by Brengelmann, 1960), as part of expressive
movements generally, to appear in the famous Handbook of Abnormal
Psychology. Later, a research student at the Institute of Psychiatry compared
EPQ scores with objective measures such as word spacing and letter height for
158 pairs of twins, but found little that was significant; for example, the mean
correlation between E scores and six measures of size was 0.06 (Stabholz,
1981). In 1986 Eysenck and Gudjonsson had a professional graphoanalyst
answer the EPQ on behalf of 99 randomly chosen adults based on their
handwriting, but the mean correlation with their actual EPQ scores was 0.05,
or definitely not useful—like the results of the dozen latest studies reviewed by
them. They concluded:
The general outcome, at least as far as this particular graphologist and this particular
sample of persons and traits is concerned, is not dissimilar to the picture obtained from
the other studies quoted. There is very little relationship between personality and
graphological analysis ... (Eysenck & Gudjonsson, 1986, p. 264)
Their study appeared in Personality and Individual Differences, which
subsequently published three more graphology studies. Furnham and Gunter
(1987) found that handwriting features and EPQ scores correlated overall
below chance level. Neter and Ben-Shakhar (1989) meta-analyzed 17 studies
 Graphology, astrology, and parapsychology 517

and found that graphologists were no better than nongraphologists in

predicting work performance from handwriting. Edwards and Armitage
(1992) found that graphologists were little better than nongraphologists in
matching handwritings to contrasting personalities or occupations.

2.4 Present status of Eysenck's conclusion

Eysenck's conclusion in 1986 was no different from his earlier ones. A later
meta-analysis by Dean (1992) of over 100 graphological studies (see Table
23.2) reached exactly the same conclusion. The correlation between
graphologists' judgments and reality is nonzero, but it is not big enough
(0.12) or reliable enough (0.42) to be useful. Other methods are better—
usually much better. The outcome was largely unaffected by study quality, with
no indication that future studies would give better results (see Figure 23.1).

Effect size vs. year of study. 2000

As methods improve over time,
so the results should converge
on the truth with a correspond-
»i .
10Jin
IWU
ing decrease in scatter. But in
°0«0 each case, despite advances in
Trend excluding
/ * I A * this outlier testing over the time spans
involved, there is no apparent I96°
• *• support for this.
1940
Effect size vs. sample size.
• Neutral content As sample size increases,
or features.
O Non-neutral or sampling errors decrease, so
uncertain content each plot should resemble an 100
inverted funnel. And they do. In
each case there are fewer
extreme results on the left than 50
on the right, suggesting that
publication bias exists.
o
Distribution of effect sizes. I
U Mean of 40
The distribution is orderly, effect sizes
weighted by
suggesting that most of the sample size
variability is due to sampling '0.050
standard
error, which agrees with the deviation
meta-analytic results. Left: The
(
bar width is too coarse to
resolve the small neutral/non-
neutral difference discernible
in the other plots. <

Figure 23.1. Visual analysis of 107 graphological effect sizes (left) and 40 astrological effect sizes
(right) from Table 23.2, using the plots suggested by Light & Pillemer (1984).
518 Further Eysenckian interests

7JUp,

IQ 70 vs. IQ 132 (Castelnuovo-Tedesco, 1948, pp. 199, 200).

ttia ia>
t/o^f
V
Artist (Paterson, 1976, p. 25) vs. chronic alcoholic (Sonnemann, 1950, p. 239).

Normal vs. blind drunk (Greene & Lewis, 1980, pp. 180-181).
Differences due to 0.1% blood alcohol are easily detected, see Rabin & Blair (1953).
I liked seeing your gardin.
Thank you for the excursion through horticultural delights.
If content is not controlled, the difference can show up even when typed (invented by us).

,/ QsTl*. O*^ f£&+&>U4+£' ±-

I CMV\

But if other differences are large, such as sex, occupation, and so on, you quickly get lost. Which
ones are lying? Answer: they all are. EPI E < 5 on left, E > 20 on right.

6-nje,
I UA<TRJU
Unsurprisingly, neat writing in school gets 15% better marks than untidy writing for the same
essays. Best and worst of ten styles by 11-year-old children (Briggs, 1970, p. 52).

Figure 23.2. If differences between people are big enough it would be unreasonable not to see
them reflected in their handwriting.

This still leaves the question of why graphological effect sizes are nonzero.
But if differences between people are big enough it would be unreasonable not
to see them reflected in handwriting (see Figure 23.2). Take the two biggest
single determinants of human behavior and destiny, namely gender and
intelligence. Many studies have shown that lay judges can pick gender from
handwriting with 60-70% accuracy (effect size 0.2 to 0.4), or even more when
 Graphology, astrology, and parapsychology 519

only their most confident judgments are counted. Similarly many studies have
shown that intelligence can be judged from handwriting with an effect size of
around 0.3 (see Dean, 1992). Figure 23.2 shows how easy it is to distinguish
between extreme IQs, between the artistry of the artist and the tremor of the
alcoholic, and between being sober and blind drunk (here we will ignore the
point that it is even easier face to face!). Subjects can also manipulate their
image; for example, by writing neatly to emphasize their supposedly orderly
mind (not a ploy ever followed by Eysenck!). That graphological effect sizes
are nonzero is therefore unremarkable. The same cannot be said for the claims
of astrology or parapsychology.

3. ASTROLOGY
3.1 Introduction
Eysenck was involved with astrology mostly during 1975-1985, a decade after
he became involved with parapsychology. It was an interesting choice. To start
with, astrology has two things in common with parapsychology, namely a solid
core of testable ideas, and no obvious explanation for its alleged effects.
Otherwise there are major differences. Astrology does not explicitly involve
psi, it is heavily tainted with the nonsense of newspaper horoscopes, and even
its serious literature (totaling some 250 shelf-meters of Western-language
books and periodicals) abounds with inconsistent and contradictory assertions.
"It would not be unfair to term this literature a nightmare" says the
parapsychologist Carl Sargent (1986, p. 350). But certain facts seemed to be
persuasive.

3.2 Eysenck's first involvement with astrology

Eysenck's interest in astrology was aroused by the findings of the late French
psychologist Michel Gauquelin (1926-1991). Over the years Gauquelin had
disconfirmed many of astrology's claims, but one curious finding persisted. At
the birth of eminent professionals, such as soldiers and painters, certain
planets tended to prefer or avoid the areas just past rise and upper
culmination, roughly one-fifth of the total. The surplus or deficit could not
be explained by demographic or astronomic artifacts and was typically 10-25%
more than expected. (This may seem like a lot but the corresponding effect
size, which for 30 years nobody bothered to calculate, was only 0.02 to 0.05.) To
add to the puzzle, the effect disappeared if the professionals were not eminent.
The usual explanations such as sampling errors did not apply because
Gauquelin used large samples with N sometimes exceeding 3000; the results
were highly significant with p often below 0.0001; and the effect seemed to
replicate. Eysenck (1975a) was impressed, and gave a favorable review of the
Gauquelin results in the magazine New Behaviour.
520 Further Eysencldan interests

In the same article Eysenck took issue with Karl Popper's argument that
astrology, like psychoanalysis, is a pseudoscience because it consists of
untestable assertions. Eysenck argued that, on the contrary, astrology does
make testable assertions such as those Unking planetary positions and
personality, hence "there should be no difficulty in arranging an experiment
to test the hypothesis quite unambiguously." In giving Gauquelin's research as
an example, Eysenck commented:
I think it may be said that, as far as objectivity of observation, statistical significance
of differences, verification of the hypothesis, and replicability are concerned, there
are few sets of data in psychology which could compete with these observations ... I
think we must admit that there is something here that requires explanation.
This is a good illustration of Eysenck's insistence on giving priority to facts
over mere opinion.
Gauquelin had shown that planetary positions also seemed to be related to
personality as determined by traits extracted from biographies. This led to a
joint study with Sybil Eysenck who translated the traits into ENP (extraversion,
neuroticism, psychoticism). The results confirmed the original findings (M.
Gauquelin, F. Gauquelin, S., Eysenck, 1979), and were subsequently replicated
using an independent sample (M. Gauquelin, F. Gauquelin, S., Eysenck, 1981).
A decade later, Ertel (1990, 1991, 1993) and others found a previously
unsuspected bias. When extracting trait words, Gauquelin had probably been
influenced by knowledge of planetary positions, a sort of experimenter effect,
which then carried through when the trait words were translated into ENP.
The planetary relation with personality was therefore probably an artifact. But
the planetary relation with occupation remained unaffected (Ertel 1992).

3.3 The Mayo sun-sign study

At the time of his New Behaviour article Eysenck was also becoming involved in
a test of sun signs. In 1971, the British astrologer Jeff Mayo had sent Eysenck a
study of 1795 subjects in which their mean scores on Mayo's extraversion
questionnaire were plotted against sun sign. The results showed a zigzag
pattern completely in accordance with astrology. Eysenck was intrigued and
made the EPI (Eysenck Personality Inventory) available to Mayo for further
tests. Then in 1973, quite independently of Mayo, the British sociologist Joe
Cooper showed Eysenck a study of Bradford University students in which their
mean E scores were plotted against sun sign. The results showed the same
zigzag pattern.
The outcome was a paper by Mayo, White, and Eysenck (1978) detailing the
EPI results for 2324 subjects, followed in the same journal by a paper by
Smithers and Cooper (1978) detailing the EPI results for 559 students. In each
case the result was a zigzag pattern in agreement with astrology. However, the
 Graphology, astrology, and parapsychology 521

difference in mean E score between odd and even signs was only 0.7, which is
very small compared to the mean E score of about 13 and SD of about 4, and is
much smaller than the claims of sun-sign astrology would suggest. In due
course this led to over a dozen replications, most of them positive, the mean
effect size being 0.09 (see Table 23.2).
Advance notice of the Eysenck paper was hailed by astrologers as "possibly
the most important development for astrology in this century" (Phenomena,
1977,1, 1). But the effect disappeared when people unfamiliar with sun-signs
were tested, so it had a simple explanation—prior knowledge of astrology. Ask
Sagittarians (who are supposed to be sociable and outgoing) whether they like
going to parties, and astrology might tip their answer in favor of yes rather than
no. The same bias applies to the other signs, who are supposed to be
alternately E+ and E-, hence the zigzag. The bias may be unconscious and
very slight but in large samples it can attain impressive significance, as in the
Mayo study where/? = 0.000005 (although reported by convention only asp =
0.0001) for N = 2324.
The mean difference of 0.7 in mean E score is equivalent to one person in
three changing their response to one E question, half in the E+ direction and
half in the E— direction. This is compatible with the various national opinion
polls which have found that around one-third of the population believe in at
least some parts of astrology. In other words it seems that roughly one person
in three not only believes in astrology but also believes in it sufficiently to
measurably shift their self-image in the corresponding direction. Whatever our
opinion of astrology, its power to affect so many people in this way deserves
recognition.
Mayo (1986, p. 232) subsequently claimed that the joint paper was mistaken,
because most of the subjects were "only vaguely interested in astrology or not
at all," so it was ridiculous to say the results were due to prior knowledge.
Eysenck (1987b) pointed out that the effect also disappeared for people who
knew too much about astrology to have faith in sun signs alone, so Mayo's
observation merely confirmed the explanation. "Mayo makes it appear as if the
effects of belief in astrology would have to be very powerful, and would have to
be much more systematic than the very slight effect necessary to produce the
outcome would have to be."
Some workers have criticized Eysenck for premature publication of his work
with Mayo, because the results turned out to rest on an artifact. But Eysenck's
defense would presumably be that "science is a self-correcting process." The
paper did provoke numerous replications, so the false nature of the original
claim was probably demonstrated all the sooner as a result. Moreover, Eysenck
had warned in the paper that a possible "weakness of the study" was that the
data were collected from people interested in astrology. When later sending
out reprints he often enclosed an accompanying note inviting the reader to
522 Further Eysenckian interests

suggest how the results might have come about. These tactics illustrate
Eysenck's faith in factual evidence and his emphasis on correct interpretation.
The same tactics were used in the case discussed next.

3.4 The Fuzeau-Braesch twin study

In 1992 an astrology study was accepted for publication in Personality and
Individual Differences, only the fifth in its history. In it the French biologist-
astrologer Suzel Fuzeau-Braesch (1992) had the parents of 238 pairs of twins
match their twins' personalities with brief astrological descriptions prepared by
her from the twins' birth charts. The result was 68.5% hits, very significantly
better than the 50% expected by chance. Eysenck (1993) gave Astrologie: La
Preuve par Deux, Fuzeau-Braesch's book-length account of the same research,
a favorable review but with due emphasis on the need for replication:
The study was read by several highly critical referees who made useful suggestions
but could not fault it in design or analysis. Does it prove that some astrological
predictions are veridical? Of course not. Replications are urgently required, and only
if these are equally successful could one come to any really positive opinion. Certain
aspects of the study might need modification, but in essence it is a proper scientific
study when evaluated by the usual criteria.
As it happened, not only did an independent replication by O'Neill give
negative results, but also a detailed appraisal by Ertel and Dean found puzzling
circumstances. Among other things, Fuzeau-Braesch's results were too good to
be true, her astrological descriptions frequently did not agree with those
prescribed by her rules, and the interval between her twin births did not
increase with increasing personality difference as required by her claims.
Worse still, there were discrepancies between her reported data and her
original data, which had been obtained with great difficulty only after direct
appeals to Eysenck. A re-analysis of the original data showed no significant
deviation from chance, which was in agreement with O'Neill's independent
replication (Ertel & Dean, 1996). As with the Mayo study, the false nature of
the original claim was probably demonstrated all the sooner by its publication.
The other four astrological studies to have appeared in Personality and
Individual Differences began with Gauquelin et al. (1981), whose successful
replication of an (artifactual) association between planets and personality has
already been mentioned. Tyson (1982) found that clients visit astrologers
mainly to reduce social and relationship stresses. Startup (1983) found no
significant difference between the personalities of astrology and psychology
students. Tyson (1984) found that subjects could not pick their own birth chart
interpretation from four others better than chance.
 Graphology, astrology, and parapsychology 523

3.5 Facilitating astrological research

Just before the Mayo study appeared, Recent Advances in Natal Astrology
(Dean et al., 1977) was published. This was the first critical scientific review of
the research basis to astrology. To ensure accuracy it involved a total of 54
collaborators, one of whom was Eysenck, who assisted with the sections on
personality and psychology. This review took seven person-years to prepare,
surveyed many hundreds of books and articles, and documented over 150
empirical studies published in astrology journals and over 20 published in
psychology journals. By 1996, new studies and previously missed studies had
brought those totals to about 300 and 70, respectively.
The availability of the Dean et al. (1977) review and the encouraging results
of the Gauquelin, Mayo, and Cooper studies prompted Eysenck to take his
interest further. He facilitated the use of the Institute of Psychiatry by
astrologers and psychologists for a joint weekend research seminar in May
1979 (Gibson, 1981, p. 209). The seminar was a success and by 1996 nine more
had been held. The first four were held at the Institute of Psychiatry, the rest
(from 1986) were held elsewhere in London. Attendance has varied from a
high of 110 in 1981 to a low of 45 in 1989, with a mean of 70, of which typically
15 are speakers.
The interest of Eysenck in astrology led to three "Eysenck Research
Seminars" being organized by astrology-minded academics at Long Beach,
Freiburg, and Naples in 1986, 1987, and 1988 respectively. The first had a
dozen speakers but attracted only 25 people, despite being timed to follow two
major US astrology conferences of the traditional kind, each of which attracted
about 1000 people. Nevertheless the seminar was significant because it resulted
in the formation of the Committee for Objective Research in Astrology
(CORA), with Eysenck as Chairman, and 12 others (astrologers and
academics) chosen for their expertise. CORA was set up to counter the
generally poor quality of astrological research by providing free guidance and
advice. But only a few inquiries are received each year and most of these do not
proceed further.

3.6 Astrology: science or superstition ?

In the late 1970s, Eysenck and Nias began a survey of the scientific evidence for
astrology aimed at a more general readership than was Recent Advances. The
result was the book Astrology: Science or Superstition? (Eysenck & Nias, 1982),
which covered astrological principles, sun-signs, marriage, illness, suicide,
appearance, time twins, season of birth, terrestrial and solar cycles, radio
propagation, earthquakes, lunar effects, and the work of the Gauquelins.
Because it addressed a complex unfamiliar field characterized by a large and
unsatisfactory literature, the text was submitted in whole or in part to a total of
nine experts for comments, which greatly improved accuracy and balance.
524 Further Eysencldan interests

Despite the availability of Recent Advances, the original literature was accessed
wherever possible, resulting in a stack of photocopies half a meter high. This
illustrates the care taken to be independent and to get the facts right. Also, new
material was discovered resulting in about 40% of the book's 230 references
being additional to those appearing in Recent Advances. In 1986, the book
appeared as a Penguin edition; it has also been translated into seven languages,
the latest being Norwegian.
Apart from reviewing the evidence for astrology, Eysenck and Nias made
two original and important refutations, both of which were subsequently
confirmed by others. First, the Mayo zodiac effect was shown in two separate
studies to be an artifact of prior knowledge. Second, the claims of John Nelson
(1951, 1978), that planetary positions can be used to predict shortwave radio
quality with about 90% accuracy, were shown to rest on an artifact in
calculating the accuracy rate—the correlation between prediction and outcome
for 5507 predictions was subsequently found by Dean (1983a, b) to be 0.01 or
almost exactly chance.
With respect to other claims, Eysenck and Nias were able to point to various
nonastrological explanations, for example much of the acceptance of
astrological readings was explained by the Barnum effect. But they remained
puzzled by the Gauquelin findings, agreeing with the now-famous quote of
Arthur Mather (1979): "Both those who are for and against astrology (in the
broadest sense) as a serious field for study recognize the importance of
Gauquelin's work. It is probably not putting it too strongly to say that
everything hangs on it."

3.7 Clarity of expression: I

Overall, the astrology book reflected Eysenck's skill hi explaining complex
issues in everyday language. As one of the quotes on the back cover puts it:
"No scientist of our time, that I have ever read, can match Hans Eysenck in
marshalling relevant data, presenting them lucidly, and drawing from them
plausible conclusions." For example, on the argument that no factor in a birth
chart shall be judged in isolation, and that judging all factors collectively is too
subtle a process to be investigated scientifically, an argument which can occupy
astrologers for several pages, it is all over hi one paragraph:
But this argument misses the point. If the most basic tenets of astrology are true, they
should be detectable in their own right, regardless of other subtleties. To take an
analogy, suppose we were investigating the belief that there is a connection between
diet and body weight. Of course many other factors come in, such as genetic make-
up, age, exercise, health, and so on. Nevertheless, if we took a large enough sample,
we should certainly expect to see indications that fat people tended to be well fed and
starving people tended to be thin. If astrology is true, it must pass that kind of test,
(p. 31)
 Graphology, astrology, and parapsychology 525

Another example, this time on tests of chart interpretations, which typically

involved 10 subjects and many astrologers, and which seemed to show weakly
positive results:
However, this is probably not the crucial test to make. The success or failure of the
experiment may have been inadvertently determined at a much earlier point, in the
selection of the subjects themselves. Given the present kinds of descriptions used in
astrology, clearly the life patterns of some subjects will agree with astrological
prediction, while those of others will disagree. If we now pick five or 10 people to
provide birth times and details about their profession or personality, then we will pick
some who are astrological confirmers, and others who are astrological disconfirmers,
that is, some for whom astrology would be judged right and others for whom it would
be judged incorrect in terms of agreement between their birth chart and their
personality or occupation. We can see how the outcome of the experiment is
predetermined to a large extent by the selection made at this stage ... Quite generally
it should be said that the number of subjects used is the crucial variable, and the
number of astrologers relatively unimportant; this is exactly the opposite to the
attitude taken by the investigators in all these studies, (pp. 86-87)
This was the first time that this particular sampling problem had been aired
in the astrological literature. Many investigators took no notice, but some did.
For example Dean (1986) gave the birth charts of 120 extreme E subjects and
120 extreme N subjects to 45 astrologers, who had to decide which extreme
each subject was. If anything they performed worse than another 45 astrologers
without birth charts who simply guessed the answer. When the sample was
divided into 12 sets of 10 birth charts each, the resulting 12 hit rates showed a
dramatic spread exactly as predicted by Eysenck and Nias.

3.8 Present status of Eysenck's conclusions

Eysenck and Nias concluded that, with the exception of Gauquelin's positive
findings, there was precious little evidence to support any of astrology's claims.
Nearly 15 years later this still applies (Dean, Mather, and Kelly, 1996) (see
Table 23.2 and Figure 23.1). Gauquelin's findings have been subject to several
replications, but in each case the (skeptical) researchers claimed the results
were negative, whereas others claimed the opposite. The disagreement in the
two most recent replications was about eminence. In both cases the researchers
found no effect for the sample as a whole, but others found that the sample was
insufficiently eminent and that the high eminence cases did indeed show a
planetary effect (Ertel & Irving, 1996). That the skeptical groups should pick
inadequate samples moved Eysenck (1996) to comment about:
... the incredible shenanigans to which the three hostile replication groups resorted
when, to their horror, results of their studies turned out favourable to Gauquelin.
These accounts really have to be read, savoured and appreciated by anyone who
believes that hard scientists are concerned with facts, truth, evidence ... Readers are
526 Further Eysenckian interests

invited to try it out! Just tell your scientific friends what the facts are. They will
squirm, put up all sorts of irrational objections, argue that the facts can't be true—
and finally refuse to look at the facts! Nothing has changed since Aristotelian
astronomers refused to look through Galileo's telescope to see the four moons of
Jupiter. This too is an interesting psychological phenomenon we might well
investigate.
This echoes the conclusions of others who reviewed the earlier skeptical
investigations (Curry, 1982), namely that scientists all too often have precisely
those irrational biases that their training is supposed to prevent. But is there a
plausible mechanism to explain Gauquelin's planetary effects? As Dean et al.
(1996) point out,
Most of the Gauquelin data comes from the nineteenth century, at a time when births
were reported by the parents, when occupations tended to run in families, when
popular almanacs featured the risings and settings of the visible planets, and the
ability to read increased with eminence. Some parents would therefore have had the
opportunity to adjust the time of birth to conform with planetary beliefs, thus
creating detectable planetary effects. As it happens the Gauquelin data show
evidence of even grosser manipulation by superstitious beliefs; for example, there is a
consistent deficit of births on the thirteenth. If days can be adjusted, hours should be
even easier, and indeed there is a consistent large deficit of births at midnight, i.e.,
the "witching hour," which is understandable given the massive witch hunts that
terrorized Europe during the sixteenth and seventeenth centuries. The same deficit
does not occur when, as in a modern hospital birth, parents are excluded from the
reporting process. Finally, the same almanacs did not feature signs or aspects, which
is consistent with Gauquelin's failure to find the corresponding effects.
If this explanation is correct, Eysenck will no doubt relish the irony of critics
strenuously denying an effect that was, like the zodiac effect, completely
nonastrological. Eysenck's contribution to this debate was his insistence that
there was an effect to be explained and his refusal to be satisfied with
dismissive explanations. Unsurprisingly, the response by astrologers has been
generally to praise him when his comments are favorable to astrology, and to
condemn him otherwise. The same ambivalence was true of phrenologists, and
if our analysis of the present status of astrology is along the right lines,
astrologers may well share their fate. But will the same also be true of
parapsychologists?

4. PARAPSYCHOLOGY
4.1 Introduction
Parapsychology was an earlier interest for Eysenck than astrology, but it had
the same attractions—testability, promise of new knowledge, and apparently
positive results. However, parapsychology has a much larger body of research
than astrology (several thousand studies vs. several hundred) and many more
 Graphology, astrology, and parapsychology 527

competent researchers. But parapsychologists understandably distance them-

selves from astrology. On the other hand, many of the alleged psi effects are
precisely those simulated by mental magicians, and many alleged psychics are
known to use trickery. Unlike astrologers, parapsychologists have acted on
criticisms, so that recent years have brought commendable advances in
procedure (notably ganzfeld and random number generators), in data analysis
(notably meta-analysis), and in uncovering fraud (notably Soal).
Rao and Palmer (1987) note that most Americans think they have
experienced one or more psychic events in their lives. So "whatever the
explanation of psychic experiences, they happen, they are common, and they
are often important to people" (p. 540). Krippner (1987) notes that, far from
being surprising, many of these reported experiences would be expected from
what we know about perception, cognition, and affect. Science seems well on
the way to providing explanations for experiences such as acupuncture, out-of-
body experiences, and near-death experiences that a few decades ago seemed
totally mysterious. Indeed, recent findings on temporal lobe activity may even
explain why some people think they have been abducted by UFOs, or have had
psychic experiences (Blackmore, 1994). But Eysenck has been less concerned
with psychic experiences than with laboratory research.
Effect sizes in parapsychology are small (see Table 23.2). So there are
problems of experimental purity that many psychologists know little about (nor
anybody else). The problem is made worse by psi being defined negatively—
"as precisely not the product of any means we can think of (Flew, 1987, p. 96).
Finally, since we have no idea how to guarantee the absence of psi, the
standard for comparison is usually a statistical model. But as pointed out by
Gilmore (1987), nothing requires that the real world should behave like a
statistical model. (He adds that apparent deviations can therefore be properly
assessed only by Monte Carlo methods, which are generally not used by
parapsychologists.) Even simple things can be impossible to decide; for
example, do you give the lowest quality rating to a procedure accompanied by
details showing it to be inadequate, or to a procedure accompanied by no
details? Readers interested in the technicalities of psi research will find no
better introduction than the nearly 50 commentaries by parapsychologists,
psychologists, neuropsychologists, physicists, philosophers, and statisticians on
the target papers by Rao & Palmer (1987) and Alcock (1987). Science at its
best.

4.2 Eysenck's early writings on parapsychology

A chapter on "Telepathy and Clairvoyance" in Sense and Nonsense in
Psychology (Eysenck, 1957) formed Eysenck's first writings on this subject. In it
he outlined the state of knowledge at that time, and concluded that psi is a
reality. He also concluded that it is not universally accepted by scientists
528 Further Eysencldan interests

because as soon as they leave the field in which they have specialized they "are
just as ordinary, pig-headed, and unreasonable as anybody else." The question
of fraud as an explanation of psi results was discounted in the now famous
quotation:
Unless there is a gigantic conspiracy involving some thirty University departments all
over the world, and several hundred highly respected scientists in various fields, many
of them originally hostile to the claims of the psychical researchers, the only
conclusion the unbiased observer can come to must be that there does exist a small
number of people who obtain knowledge existing either in other people's minds, or in
the outer world, by means as yet unknown to science, (pp. 131-132)
Ten years later Eysenck (1967) tried to add a theoretical basis to the
empirical evidence for psi. He responded to the suggestion of Rao (1966) that
"there is no intrinsic reason why personality differences should help or hinder
psi if it is like other abilities such as perception or memory." Eysenck argued
that, on the contrary, if psi is a primitive form of perception, evolving before
higher forms of perception based on the cortex, then cortical arousal and
associated traits of introversion should hinder the ability. Psi would thus be the
opposite of the other senses, and so be truly etfra-sensory! He pointed out that
(1) psi scores tend to decline during test sessions, suggesting a monotony
factor; (2) introducing novelty in testing sessions helps; and (3) subjects
typically perform better under spontaneous rather than rigidly controlled
conditions, all of which is consistent with extraverts performing better than
introverts. He then provided support for this from a survey of the limited
literature at the time.
Furthermore, Eysenck argued that the association with extraversion
constitutes evidence against faking because presumably "the investigators did
not know of the hypothesis in question." But here Eysenck had forgotten that
the propensity for practical jokes also increases with increasing extraversion!
Obvious questions such as how many tricksters among the subjects would be
needed to produce the same results ought to have been addressed but were
ignored. The point is not that contrary views are right or wrong, but that
Eysenck's viewpoint was too one-sided in an area which could least afford it.
Eysenck also made an important methodological point, namely that psi
researchers had not followed the standard practice of calculating reliabilities.
Such a practice guards against subjects who consistently score above chance
being counter-balanced by others who consistently score below chance, a
situation which is suggested by his argument for personality effects. Hence
individual reliabilities "should always be calculated as a matter of routine."
However, for some reason most psi researchers have not taken this advice.
The fifteenth edition of Encyclopedia Britannica includes a review by
Eysenck (1974) on psi. Here he points out that "the very existence of
parapsychological phenomena is still very much in dispute," probably because
 Graphology, astrology, and parapsychology 529

critics fail to present arguments "supported by a survey of all the known facts."
The common practice of citing isolated studies is not enough—it is the balance
of evidence that must be considered. After describing how psi experiments
have been investigated by the American Psychological Association and the
American Statistical Society, without any adverse criticisms being made,
Eysenck concluded that "the evidence for ESP is stronger than that for many
tenuously supported psychological phenomena." Interestingly, in later
printings of the fifteenth edition, Eysenck's review was eventually replaced
by generally negative articles almost as if his views were seen as dangerously
heretical. For example, the 1989 printing contains the unqualified claim that
"many scientists continue to doubt the existence of ESP" (4, p. 638).

4.3 The precognition experiment

In an attempt to "highlight certain methodological and theoretical considera-
tions which have hitherto played very little part in parapsychological work,"
Eysenck (1975b) tested the precognitive ability of rats by seeing if they could
anticipate randomly generated electric shocks. The results seemed to show that
rats did have precognitive ability. But because the equipment had a tendency to
malfunction, Eysenck organized a more extensive experiment using more
reliable equipment (Hewitt, Fulker, & Eysenck, 1978). Unfortunately, none of
the results was significant, and it was concluded that the earlier results were
spurious. While raising the question of whether Eysenck was wise to have
published the first result, this episode (as with the Mayo study) again illustrates
the power of science in being a "self-correcting process."

4.4 Explaining the unexplained: Mysteries of the paranormal

The first edition of this book (Eysenck & Sargent, 1982) was written at the
same time as Astrology: Science or Superstition?, and was followed 11 years later
by an updated second edition (Eysenck & Sargent, 1993), both with full-color
illustrations on nearly every page. By 1995 the latter had been reprinted four
times. Both editions covered altered states of consciousness, card guessing,
dreams, faith-healing, ganzfeld research, mediums, metal bending, picture
guessing, poltergeists, psychokinesis, reincarnation, psi in everyday life, and the
effects of hypnosis, personality, and meditation. The second edition included a
new final chapter on "Psi, science, and the future." Unlike the astrology book,
this one did not involve consultation with outside experts, which as we shall see
led to one-sided selection of evidence.
In their overview of the positive evidence for psi, Eysenck and Sargent (1982,
1993) attempt to show that psi effects conform to recognizable laws and
interact with variables like personality in ways that make sense. In particular
they show how a good theory can begin to make sense out of apparent disorder
530 Further Eysenckian interests

and lead to new tests. In short, the emphasis of their book is on various
theories and how they stand up to testing—a sure sign of the Eysenck
approach. Because the various theories are testable, nobody who reads their
account can fail to be impressed by the progress being made.
For example, Eysenck and Sargent (1993, pp. 76,128) discuss ESP effects in
terms of two theoretical models. In Honorton's sensory noise-reduction model,
the key is relaxation and a constant low-level sensory input. The reduction in
noise increases sensitivity to ESP after 15-20 min of habituation to these
conditions. But if the subject is having sensory inputs, ESP is unlikely to
appear. In Stanford's bias-and-rigidity-reduction model, the key is the absence
of preoccupations and constraints. ESP can then be triggered by need; for
example, to avoid a not-consciously recognized hazard, but not by too much
need, which produces stress and impaired performance. If the subject is
preoccupied and constrained, ESP is unlikely to appear. Interestingly, both
models suggest that the process of reading handwriting or astrological charts is
not conducive to ESP, even though some practitioners claim that ESP is
important for the successful translation of isolated features into a meaningful
whole.
Similarly, in Walker's quantum mechanical model of psi, the key is
randomness. The random processes in the brain can, at the quantum level,
affect randomness elsewhere. In other words no randomness, no psi. So
dynamic systems such as dice tumbling and random number generation, with
their inherent randomness, should be easier to influence than static systems
such as a delicately suspended needle (and, we might add, levitating
mediums!). The model is sufficiently mind-boggling to tax even Eysenck's
writing skills, but again the emphasis is on the testability of its equally mind-
boggling predictions. It has
... the classical scientific merit of [successfully] predicting entirely new and hitherto
unsuspected experimental results ... what is truly important here is something almost
unthinkable not so very long ago: precise mathematical statements about how psi
should operate (p. 147)

4.5 Clarity of expression: II

Mind-boggling models apart, Eysenck's legendary clarity shines through. For
example, psychic readers are sympathetically covered in just one paragraph,
with echoes of when Eysenck was himself a (supposed) handwriting reader
nearly 60 years previously:
The vast majority of mental mediums are inoffensive people whose sincerity it would
be uncharitable to question, but who produce a great deal of waffle which no
seasoned researcher takes seriously. Their services can be comforting to the relatives
of a recently deceased person and many of them charge no fee, or only a token fee.
They are not commercially oriented charlatans exploiting the gullible for profit. On
 Graphology, astrology, and parapsychology 531

the contrary, many are kindly and well-meaning. This is not to say that anything more
than self-deception is involved in their performances. It does not take much practice
to learn "cold reading" (to give quite an effective performance as a "medium") with
nothing paranormal being involved at all... Educated guesswork and following leads
from fished-for information can produce an extremely convincing piece of
"mediumship" so far as the sitter is concerned. (1993, p. 163)

4.6 Selection of evidence

Ironically, having criticized critics for being selective with the evidence,
Eysenck and Sargent (1982,1993) tend to follow the same path. For example,
neither edition mentions that in 1966 nearly every leading parapsychologist
attacked the suggestion that Seal's experiments were fraudulent. (In Eysenck's
Sense and Nonsense in Psychology, Seal's work is described as the "most
impressive perhaps of all the studies in precognition," p. 136.) But in 1978,
fraud was established beyond doubt. Similarly, in 1984 the Parapsychological
Association set up a committee to investigate possible irregularities in
Sargent's ganzfeld studies, but he repeatedly refused them access to his raw
data (Blackmore, 1996). Such a refusal is of course the ultimate scientific sin,
and is an awkward behavior to find in your co-author. Yet this episode is
ignored by Eysenck and Sargent.
Again, both editions claim that the famous physical medium Daniel Dunglas
Home was never caught red-handed in the act of fraud. Stein (1993) claims this
is untrue—Home was never caught in public but he was allegedly caught
several times in private sittings. Furthermore, the magician Milbourne
Christopher (1970) has plausible explanations for many of Home's
phenomena and has managed to duplicate Home's more baffling feats
including levitations. The second edition, unlike the first, does list some
skeptical accounts of Home but dismisses them as "a sorry collection" (p. 28).
Similarly it fails to deal with Alcock's (1990) detailed critique of Schmidt's
work with random number generators, which runs to 16 pages plus a 45-page
study-by-study assessment of Schmidt's work (for a summary see French,
1996). Finally, Eysenck and Sargent give no hint of the various prizes offered
for convincing evidence of psi, ranging from the late Joseph Dunninger's
$10,000 and LeBon's $85,000 to the various offers by skeptic groups around the
world then totaling about half a million dollars (and considerably more today).
It says a lot that nobody has succeeded in winning one of these prizes. Sargent's
(1986) reply to this criticism was to imply that the offers were rigged.
The same one-sidedness is evident in Eysenck and Sargent's attitude to
critics. One of the few critics to receive their praise (albeit not without very
strong criticism as well) is Ray Hyman, whose cooperation with para-
psychologist Charles Honorton led directly to the development of the auto-
ganzfeld technique. But there are literally hundreds of instances where critics
532 Further Eysenctdan interests

have shown that allegedly paranormal events could be explained completely in

nonparanormal terms. James Randi, for example, has exposed numerous fake
psychics, including faith healers who were making huge sums by exploiting the
sick and the needy. Surely Eysenck and Sargent should have recognized that
even extreme critics such as Randi are not only providing a vital public service
but are also actually helping serious parapsychologists by exposing such
charlatans? There can be little doubt that most of what is generally perceived
to be paranormal is nothing of the kind (French, 1992).
Of course in any popular book there will be constraints on what can be
included. No doubt there will also be pressure from the publisher to avoid
undue skepticism (skeptical books are not what the public wants and therefore
do not sell). As a refreshingly open-minded survey of many controversial
claims the book can only be welcomed. But as before, the point is not that
contrary views are right or wrong but that Eysenck and Sargent's viewpoint
tends to be one-sided in an area which can least afford it.

4.7 Know Your Own Psi-Q

The need for a balanced view is even more apparent in Eysenck and Sargent's
(1984) second book, Know Your Own Psi-Q. The objection of skeptics is
perhaps best illustrated by Randi (1984). He describes the book as
a disaster in every way except one: it may provide us with an accurate picture of just
how naive the authors are in designing proper protocol for testing psi-powers. If their
book correctly expresses the standards of parapsychologists in general, it is no
wonder that the rest of the scientific community scoffs at their efforts.
Randi points out that the two-and-a-half pages of their bibliography contain
not a single skeptical work. He concludes that their book gives "a totally one-
sided view of the subject." Sargent (1986, pp. 372-373) dismisses Randi's
comments as "presumably willful and for polemical purposes." He points out
that the book was not meant as a "manual of laboratory research methods,"
but then contradicts this by saying it is "a manual of method" (his emphasis).

4.8 Fraud, credulity, and methodological errors

In his early writings Eysenck consistently warned about trickery in
parapsychology.
A few hours' instruction in elementary conjuring should enable any reasonably adept
person to produce most of the alleged psychical phenomena seen at seances.
(Eysenck, 1957, p. 113)
Investigators who cannot explain every trick performed by stage magicians should
consider themselves barred from investigating alleged psi phenomena. (Eysenck,
1974)
 Graphology, astrology, and parapsychology 533

By contrast, trickery and fraud were rarely considered in Eysenck and

Sargent's first edition. However, they are discussed several times in the second
edition, albeit often only to be dismissed as possible explanations. Here
Eysenck and Sargent (1993) repeat Eysenck's (1983) earlier sentiments that
"experiments in parapsychology are at least as rigorous as most of those
published in psychological journals in more 'reputable fields', and probably
more so." But this misses an important point—parapsychology has far more
revolutionary implications than psychology, and therefore demands a
correspondingly higher standard of evidence. Also, surveys of fraud in science
generally have shown it to be far from negligible; even the mere incidence of
errors is "rather large" (Kohn, 1986, p. 207). Furthermore, as discussed next,
the supposed rigor may sometimes be illusory.
People who read about magic and conjuring may well see it as a challenge to
see if they can fool an academic researcher. The books they read often describe
psychic ability in terms of clever trickery. For example, Fulves (1979) describes
the last of his 67 mind-reading tricks as "a staggering demonstration of
paranormal ability, an overwhelmingly positive test that ESP exists." Similarly,
in a reference book on mental magic intended for both magicians and
parapsychologists, Kaye (1975) includes an annotated bibliography of 120
books and articles on mental magic, and explains dozens of tricks that to the
uninitiated are totally baffling. One would therefore expect researchers to be
aware of mentalist techniques, and to involve magicians when designing their
studies.
Unfortunately, quite the opposite situation often prevails, as was
demonstrated by the magician James Randi who persuaded two fellow
magicians to pose as psychics and infiltrate the psychical research laboratory
newly established by a $500,000 grant from the McDonnell Foundation.
According to Randi (1983), during three years of testing, the researchers were
fully persuaded that the magicians had genuine psychic powers. They also
ignored all the precautions that Randi had suggested, and for two years
continually rejected his offers of help. According to Thalbourne (1995), the
account by Randi was one-sided, some of the actions by the magicians (like
leaving windows open to allow later unauthorized entry) bordered on the
criminal, and the way the hoax was revealed (by a full media circus without
inviting the researchers to present their side of the story) was repugnant to
many. Nevertheless the hoax has generally persuaded parapsychologists to
collaborate more closely with magicians.
In addition to deliberate fraud and trickery, methodological errors may also
produce results which appear to support the existence of psi. For example, in a
survey of necessary precautions, Schmeidler (1977, p. 133) comments "It is well
known that if light strikes the back of an ESP card at the right angle, it is
possible to read the symbol through the back of the card." Because psi can be
534 Further Eysenckian interests

evaluated only as departures from a statistical model, and because departures

can have many non-psi causes, the detection and control of methodological
errors is absolutely crucial.
Then there is publication bias or selective reporting, which is visible for
graphology and astrology in Figure 23.1. Eysenck and Sargent (1993) state that
"selective reporting is not really a problem in parapsychology" (p. 20). But it
was once. Only 20 years ago Greenwald (1975) could describe parapsychology
as being so plagued by publication bias that "no reasonable person can regard
himself as having an adequate basis for a true-false conclusion." But in 1975
the Parapsychological Association passed a resolution requiring all affiliated
journals to publish strictly on merit regardless of outcome. This does not
guarantee that researchers will still submit negative findings for publication, or
that they will be accepted, but it was still a useful advance. Today, critics
generally accept that to explain away the published positive studies as
publication bias would require an unreasonably large number of unpublished
negative studies; for example, to explain away the result for random number
generators in Table 23.2 would need 10 researchers each withholding over 30
nonsignificant studies every year for 15 years! (Radin & Nelson, 1989).

4.9 Present status of Eysenck's conclusions

Eysenck and Sargent (1993) conclude
Parapsychology is a science because it shows development of theory and method. It
shows the progressive problem shifts which the philosopher of science Imre Lakatos has
persuasively argued are the hallmarks of scientific thought... it needs the involvement of open-
minded young scientists who are willing to address the evidence, and the issues, and conduct
research of their own. (p. 186)
On psi itself, they conclude that parapsychology has "certainly found"
worthwhile evidence of human capacities at present poorly understood. They
close by claiming that "parapsychology is rapidly coming of age. Despite
chronic financial undernourishment, its future as we head into a new century is
exciting. It is time that more scientists took its findings seriously, and looked at
the evidence on offer, because the best of it compares with the very best in any
area they might be working in" (p. 187). At first sight their enthusiasm does
seem to be justified.
Several current approaches look extremely promising including the auto-
ganzfeld studies (see Figure 23.3), remote viewing, remote staring (the ability
to tell if someone unseen is staring at you), and studies with random number
generators. Well-respected experts from outside parapsychology (i.e., in
addition to Eysenck) such as Robert Rosenthal, Daryl Bern, and Jessica Utts
have written very positively about parapsychology. For example, in what for
psychologists is currently the most accessible and most readable review of the
latest research, Bern & Honorton (1994) state
 Graphology, astrology, and parapsychology 535

We believe that the replication rates and effect sizes achieved by one particular
experimental method, the ganzfeld procedure, are now sufficient to warrant bringing
this body of data to the attention of the wider psychological community.
Furthermore, critics on the inside such as Ray Hyman and James Alcock
have admitted that there is something there in need of explanation (specifically
the auto-ganzfeld studies) although they have stopped short of actually
endorsing the existence of psi. For example, Hyman (1996) comments that the
latest results do not "appear to be accounted for by multiple testing, file-
drawer distortions, inappropriate statistical testing or other misuse of statistical

Figure 23.3. Ganzfeld experiments.

The subject lies down, eyes covered with halved ping-pong balls dimly lit with red or orange light,
and hears a neutral sound like rainfall through earphones. After 15-20 minutes the brain stops
attending to these unvarying sensory inputs and attends to internal nental events instead. The
subject is then ready to receive impressions from a sender In another room, who is watching a
picture or video selected at random from a set of (usually) four. The subject reports the
impressions into a tape recorder, and afterwards, on being presented with the targets, attempts to
pick the one used.
The experience is remarkably pleasant. Pedler, a double doctorate in medicine and retinal
disease who acted as a subject in Sargent's laboratory for a British TV documentary, describes it as
follows: "my mind was finally quiet; anxiety had receded, and I lay quite relaxed, waiting. It is very
difficult to describe the pictures that came into my mind, even though I spent twelve years of my life
studying the eye and vision. They were halfway between actual images, mind pictures and language
description." For example, "a black saw-tooth pattern like one of those old wood saws ... village
roof lines and chimneys ... palm trees, halfway up a mountain in silhouette" (Pedler, 1981, pp. 60-
64). Subsequently the actual target picture (people in witch's hats cutting a cake) was ranked
second. So it was a near miss.
In an awto-ganzfeld experiment the procedure is rigidly specified, with computer control of target
pictures and videos, soundproofed rooms, blind judging, and automated data storage. For 355
sessions recorded in 11 separate studies with a total of 241 subjects, the mean hit rate was 34.4% vs.
25% expected by chance, kappa = 0.12 (see Table 23.1). The highest mean hit rate was an
astonishing 50%, kappa = 0.33, recorded by 20 performing arts students, which is in accordance
with the expectation that hit rate should increase with creativity or artistic ability (from Bern &
Honorton, 1994). But Wiseman et al. (1996) point out that flaws may still exist; e.g. performing arts
students when acting as senders might well be those who would shout when becoming emotionally
involved, as senders were encouraged to be, thus posing questions about sensory leakage from
sender to experimenter, who might then influence the subject.
536 Further Eysenckian interests

inference ... Having accepted the existence of non-chance effects, the focus is
now upon whether these effects have normal causes." The last point is crucial
and, given the history of parapsychology, completely reasonable. Ever since
research began in the 1850s, parapsychology has always been on the verge of a
breakthrough that never materialized. What one generation considered a solid
case was abandoned by the next as artifacts were discovered. The same applies
in psychology—it is easy to look back at studies done 30 years ago and find
sources of artifacts that were completely unsuspected at the time.
There is another reason for adopting a more cautious enthusiasm. The
Cottingley fairies stood for over 60 years before they crumbled. It took 40 years
for the Fox sisters to admit their communications with the spirit world were a
schoolgirl prank that got out of hand. Seal's "watertight evidence" was only
revealed as definitely fraudulent some 40 years later. It took nearly 60 years to
learn that the Loch Ness Monster in the famous 1934 photo was really a toy
submarine fitted with a serpent's head made of plastic wood. Gauquelin's
planetary effect has resisted nonastrological explanation for around 40 years,
but may be on the verge of explanation. Even nonparanormal areas such as
direction-finding in pigeons, or the link between smoking and cancer, or the
existence of gravitational waves, are taking several decades to investigate and
unravel. Clearly, some mysteries can take a long time to be properly explained.

5. CONCLUSION

To end our chapter, three points might be emphasized. (1) It is easy to be

impressed by statistical significance. But if sample sizes are large, the
corresponding effect sizes can be tiny. (2) It does not take much in the way
of flaws (errors, biases, etc.) to produce tiny effect sizes. It is arguable whether
we have enough experience of flaws at this level to be confident of their
absence. (3) Only if the effect is repeatable do we have something that can be
properly investigated, at which point the mysteries can be expected to yield to
Eysenck's open-minded strategies.
By way of summary we predict that graphology will continue to show small
effect sizes for unremarkable reasons, but too small to be useful. Astrology will
continue to show effect sizes entirely explainable by nonastrological effects.
Parapsychology is the interesting one. If pressed, we would predict that the
findings, however unusual, will eventually have ordinary explanations. But even
if the explanations are extraordinary, the inevitable consequence would be a
redefinition of science to accommodate the new findings. In general we can
only agree with the observations of Rao and Palmer (1987):
It seems to us that too many commentators on both sides of the psi controversy place
excessive faith in what amounts to little more than speculations about the true nature
of the [psi] anomalies. Only by continued research, preferably supported in a
 Graphology, astrology, and parapsychology 537

meaningful way by the scientific community at large, will the speculations turn into
knowledge, (p. 551)
In each case, whatever the actual outcome, Eysenck will have played an
important role in speeding the search after some of the most elusive truths in
science. He has set us an example of open-mindedness and good science that
everyone can aim for but few will achieve.

REFERENCES

Alcock, J. E. (1987). Parapsychology: Science of the anomalous or search for the soul?
Behavioral and Brain Sciences, 10, 553-565.
Alcock, J. E. (1990). Science and supemature: A critical appraisal of parapsychology.
Amherst, NY: Prometheus.
Allport, G. W. & Vernon, P. E. (1933). Studies in expressive movement. New York:
Macmillan.
Bern, D. J. & Honorton, C. (1994). Does psi exist? Replicable evidence for an
anomalous process of information transfer. Psychological Bulletin, 115, 4-18.
Blackmore, S. (1994). Alien abduction: The inside story. New Scientist, 1952, 19
November, 29-31.
Blackmore, S. (1996). In search of the light: The adventures of a parapsychologist.
Amherst, NY: Prometheus.
Brengelmann, J. C. (1960). Expressive movements and abnormal behaviour. In
Eysenck, H. J. (Ed.), Handbook of abnormal psychology (pp. 62-107). London:
Pitman.
Briggs, D. (1970). The influence of handwriting on assessment. Educational Research,
13, 50-55.
Castelnuovo-Tedesco, P. (1948). A study of the relationship between handwriting and
personality variables. Genetic Psychology Monographs, 37, 167-220.
Christopher, M. (1970). ESP, Seers and Psychics. New York: Crowell.
Cicchetti, D. V. (1987). Differentiating between the statistical and substantive
significance of ESP phenomena: delta, kappa, psi, phi, or it's not all Greek to me.
Behavioral and Brain Science, 10, 577-581.
Curry, P. (1982). Research on the Mars effect. Zetetic Scholar, 9, 34-53.
Dean, G. A. (1983a). Short-wave radio propagation non-correlation with planetary
positions. Correlation, 3(1), 4-37.
Dean, G. A. (1983b). Forecasting radio quality by the planets. Skeptical Inquirer, 8(1),
48-56.
Dean, G. (1986). Can astrology predict E and N? 3. Discussion and further research.
Correlation, 6(2), 7-52.
Dean, G. A. (1992). The bottom line: Effect size. In B. L Beyerstein. & D. F. Beyerstein
(Eds.), The write stuff: Evaluations of graphology (pp. 269-341). Buffalo, NY:
Prometheus.
Dean, G. A., Mather, A. C. M. et al. (1977). Recent advances in natal astrology: A critical
review 1900-1976. Perth, Australia: Analogic.
538 Further Eysenckian interests

Dean, G., Mather, A., & Kelly, I. W. (1996). Astrology. In G. Stein (Ed.), The
encyclopedia of the paranormal (pp. 47-99). Amherst, NY: Prometheus.
Dwyer, T. (1986). Editorial comment. Astrological Journal, 28(3), 114.
Edwards, A. G. & Armitage, P. (1992). An experiment to test the discriminating ability
of graphologists. Personality and Individual Differences, 13, 69-74.
Ertel, S. (1990). Gauquelin contentions scrutinized. In G. J. M. Tomassen, W. de Graaf,
A. A. Knoop, & Hengeveld, R. (Eds.), Geo-cosmic relations: The Earth and its
macro-environment. Proceedings of the First International Congress on Geo-cosmic
Relations, Amsterdam, 19-22 April 1989 (pp. 255-271). Wageningen, The Nether-
lands: Pudoc.
Ertel, S. (1991). Scrutinising Gauquelin's character trait hypothesis once again.
Correlation, 10(2), 3-19.
Ertel, S. (1992). Update on the "Mars effect." Skeptical Inquirer, 16, 150-160.
Ertel, S. (1993). Why the character trait hypothesis still fails: Resolving the conflict
between the Fresno and Gottingen studies. Correlation, 12(1), 2-9.
Ertel, S. & Dean, G. (1996). Are personality differences between twins predicted by
astrology? Personality and Individual Differences, 21, 449-454.
Ertel, S. & Irving, K. (1996). The tenacious Mars effect. London: Urania Trust.
Eysenck, H. J. (1945). Graphological analysis and psychiatry: An experimental study.
British Journal of Psychiatry, 35, 70-S1.
Eysenck, H. J. (1948). Neuroticism and handwriting. Journal of Abnormal and Social
Psychology, 43, 94-96.
Eysenck, H. J. (1957). Sense and nonsense in psychology. Harmondsworth, U.K.:
Penguin.
Eysenck, H. J. (1967). Personality and extrasensory perception. Journal of the Society for
Psychical Research, 44, 55-71.
Eysenck, H. J. (1974). Theories of parapsychological phenomena. Encyclopedia
Britannica (15th edn.). Chicago: Helen Benton.
Eysenck, H. J. (1975a). Planets, stars and personality. New Behaviour, 29 May, 246-249.
Eysenck, H. J. (1975b). Precognition in rats. Journal of Parapsychology, 39, 222-227.
Eysenck, H. J. (1978). An exercise in mega-silliness. American Psychologist, 33, 517.
Eysenck, H. J. (1983). Parapsychology: Status and prospects (The J. B. Rhine Lecture).
In W. G. Roll, J. Beloff, & R. A. White (Eds.), Research in Parapsychology.
Metuchen, NJ: Scarecrow.
Eysenck, H. J. (1986). Consensus and controversy: Two types of science. In S. Modgil &
C. Modgil (Eds.), Hans Eysenck: Consensus and controversy. Essays in honour of Hans
Eysenck (pp. 375-398). Lewes, U.K.: Palmer.
Eysenck, H. J. (1987a). Anomalous phenomena and orthodox science. Behavioral and
Brain Sciences, 10, 584-585.
Eysenck, H. J. (1987b). A fair deal for astrology? (Letter). Astrological Journal, 29(2),
91.
Eysenck, H. J. (1990). Rebel with a cause: The autobiography of Hans Eysenck. London:
Allen.
Eysenck, H. J. (1993). Review of "Astrologie: La Preuve Par Deux" [Astrology: Proof by
Twins] by Suzel Fuzeau-Braesch. Personality and Individual Differences, 15, 483.
Eysenck, H. J. (1994). Meta-analysis and its problems. British Medical Journal, 309,789-
792.
 Graphology, astrology, and parapsychology 539

Eysenck, H. J. (1996). Review of "The tenacious Mars effect" (Ertel & Irving, 1996).
Correlation, 15(1), 53-54.
Eysenck, H. J. & Gudjonsson, G. (1986). An empirical study of the validity of
handwriting analysis. Personality and Individual Differences, 7, 263-264.
Eysenck, H. J. & Nias, D. K. B. (1982). Astrology: Science or superstition? London:
Temple Smith.
Eysenck, H. J. & Sargent, C. L. (1982). Explaining the unexplained: Mysteries of the
paranormal. London: Weidenfeld and Nicolson.
Eysenck, H. J. & Sargent, C. L. (1984). Know your own Psi-Q. New York: World
Almanac.
Eysenck, H. J. & Sargent, C. L. (1993). Explaining the unexplained: Mysteries of the
paranormal (2nd edn.). London: Prion.
Flew, A. (1987). Analyzing the concepts of parapsychology. In A. Flew. (Ed.), Readings
in the philosophical problems of parapsychology (pp. 87-106). Buffalo, NY:
Prometheus.
French, C. C. (1992). Factors underlying belief in the paranormal: Do sheep and goats
think differently? The Psychologist, 5, 295-299.
French, C. C. (1996). Psychokinesis. In G. Stein (Ed.), The encyclopedia of the
paranormal (pp. 604-613). Amherst, NY: Prometheus.
Fulves, K. (1979). Self-working mental magic: 67 Foolproof mind-reading tricks. New
York: Dover.
Furnham, A. & Gunter, B. (1987). Graphology and personality: Another failure to
validate graphological analysis. Personality and Individual Differences, 8, 433^435.
Fuzeau-Braesch, S. (1992). An empirical study of an astrological hypothesis in a twin
population. Personality and Individual Differences, 13, 1135-1144.
Gauquelin, M, Gauquelin, F., & Eysenck, S. B. G. (1979). Personality and position of
the planets a birth: An empirical study. British Journal of Social and Clinical
Psychology, 18, 71-75.
Gauquelin, M., Gauquelin, F., & Eysenck, S. B. G. (1981). Eysenck's personality
analysis and position of the planets at birth: A replication on American subjects.
Personality and Individual Differences, 2, 346-350.
Gibson, H. B. (1981). Hans Eysenck: The man and his work. London: Peter Owen.
Gibson, H. B. (1986). Introductory chapter. In S. Modgil & C. Modgil (Eds.), Hans
Eysenck: Consensus and controversy. Essays in honour of Hans Eysenck (pp. 3-12).
Lewes, U.K.: Palmer
Gilmore, J. B. (1987). Axioms in science, classical statistics, and parapsychological
research. Behavioral and Brain Sciences, 10, 588-590.
Glymour, C. (1987). ESP and the big stuff. Behavioral and Brain Sciences 10, 590.
Greene, J. & Lewis, D. (1980). The hidden language of your handwriting (p. 252).
London: Souvenir Press.
Greenwald, A. G. (1975). Consequences of prejudice against the null hypothesis.
Psychological Bulletin, 82, 1-20.
Hewitt, J. K., Fulker, D. W., & Eysenck H. J. (1978). Effect of strain and level of shock
on the behaviour of rats in psi experiments. Psychological Reports, 42, 1103-1108.
Hyman, R. (1996). Evaluation of a program on anomalous mental phenomena. Journal
of Scientific Exploration, 10, 31-58.
Kaye, M. (1975). The handbook of mental magic. New York: Stein and Day.
540 Further Eysenckian interests

Kohn, A. (1986). False prophets: Fraud and error in science and medicine. Oxford: Basil
Blackwell.
Krippner, S. (1987). Never say never again: Rapprochement may be nearer than you
think! Behavioral and Brain Sciences 10, 595-597.
Light, R. J. & Pillemer, D. B. (1984). Summing up: The science of reviewing research.
Cambridge, MA: Harvard University Press.
Mather, A. (1979). Response to reviews of Recent Advances. Zetetic Scholar, 3/4, 96.
Mayo, J., White, O. & Eysenck, H. J. (1978). An empirical study of the relation between
astrological factors and personality. Journal of Social Psychology, 105, 229-236.
Mayo J (1986). A fair deal for astrology? Astrological Journal, 28(5), 230-233 and 28(6),
265-268.
Modgil, S. & Modgil, C. (Eds.). (1986). Hans Eysenck: Consensus and controversy.
Essays in honour of Hans Eysenck. Lewes, U.K.: Palmer.
Nelson, J. H. (1951). Short-wave radio propagation correlation with planetary positions.
RCA Review, March, 26-34.
Nelson, J. H. (1978). The propagation wizard's handbook. Peterborough, NH: 73 Inc.
Neter, E. & Ben-Shakhar, G. (1989). The predictive validity of graphological inferences:
A meta-analytic approach. Personality and Individual Differences, 10, 737-745.
Nias, D. K. B. & Dean, G. A. (1986). Astrology and parapsychology. In S. Modgil & C.
Modgil (Eds.), Hans Eysenck: Consensus and controversy. Essays in honour of Hans
Eysenck (pp. 357-371). Lewes, U.K.: Palmer.
Paterson, J. (1976). Interpreting handwriting. London: Macmillan.
Pedler, K. (1981). Mind over matter. London: Thames Methuen.
Rabin, A. & Blair, H. (1953). The effects of alcohol on handwriting. Journal of Clinical
Psychology, 9, 284-287.
Radin, D. I. & Nelson, R. D. (1989). Evidence for consciousness-related anomalies in
random physical systems. Journal of Scientific Exploration, 5, 61-83.
Randi, J. (1983). The Project Alpha experiment. Skeptical Inquirer, 7(4), 24-33; 8(7),
36-45; and 8(2), 102-103.
Randi, J. (1984). Mirror on naive protocol (review of "Know Your Own Psi-Q").
Skeptical Inquirer, 8(2), 180-182.
Rao, K. R. (1966). Experimental parapsychology. Springfield, IL: Thomas.
Rao, K. R. & Palmer, J. (1987). The anomaly called psi: Recent research and criticism.
Behavioral and Brain Sciences, 10, 539-551.
Rosenthal, R. & Rubin, D. B. (1982). A simple general purpose display of magnitude of
experimental effect. Journal of Educational Psychology, 74, 166-169.
Sargent, C. (1986). Parapsychology and astrology. In S. Modgil & C. Modgil (Eds.),
Hans Eysenck: Consensus and controversy. Essays in honour of Hans Eysenck (pp. 339-
355). Lewes, U.K.:Falmer.
Schmeidler, G. R. (1977). Methods for controlled research on ESP and PK. In B. B.
Wolman (Ed.), Handbook of parapsychology. New York: Van Nostrand Reinhold.
Smithers, A. G. & Cooper, H. J. (1978). Personality and season of birth. Journal of
Social Psychology, 105, 237-241.
Sonnemann, U. (1950). Handwriting analysis as a psychodiagnostic tool. London: Allen
& Unwin.
 Graphology, astrology, and parapsychology 541

Stabholz, M. S. (1981). Individual differences in the handwriting of monozygotic and

dizygotic twins in relation to personality and genetic factors. Unpublished master's
thesis, Institute of Psychiatry, University of London.
Startup, M. (1983). Belief in astrology: A system of maladjustment? Personality and
Individual Differences, 4, 343-345.
Stein, G. (1993). The sorcerer of kings: The case of Daniel Dunglas Home and William
Crookes. Amherst, NY: Prometheus.
Thalbourne, M. A. (1995). Science versus showmanship: A history of the Randi hoax.
Journal of the American Society for Psychical Research, 89, 344-366.
Tyson, G. A. (1982). People who consult astrologers: A profile. Personality and
Individual Differences, 3, 119-126.
Tyson, G. A. (1984). An empirical test of the astrological theory of personality.
Personality and Individual Differences, 5, 247-250.
Utts, J. (1991). Replication and meta-analysis in parapsychology. Statistical Science,
6(4), 363-378.
Wiseman, R., Smith, M., & Kornbrot, D. (1996). Exploring possible sender-to-
experimenter acoustic leakage in the PRL auto-ganzfeld experiments. Journal of
Parapsychology, 60, 97-128.
Wolfson, R. (1951). Graphology. In H. H. Anderson & G. L. Anderson (Eds.), An
introduction toprojective techniques and other devices for understanding the dynamics of
human behavior (pp. 416-456). New York: Prentice-Hall.

Bibliography of further articles by Hans Eysenck

Eysenck, H. J. (1972). Personality and extrasensory perception. In R. Van Over (Ed.)
Psychology and extrasensory perception. New York: New American Library.
Eysenck, H. J. (1978). On Jerome's Astrology Disproved. Phenomena, 2(2), 9-11.
Eysenck, H. J. (1979). Biography in the service of science: A look at astrology.
Biography, 2(1), 25-34.
Eysenck, H. J. (1979). Review of "Recent Advances in Natal Astrology." Zetetic
Scholar, 3/4, 85-86.
Eysenck, H. J. (1979). Astrology: Science or superstition? Encounter, December, 85-90.
Eysenck, H. J. (1980). Telepathy: Sense or nonsense? Bethlem and Maudsley Gazette,
Spring.
Eysenck, H. J. (1981). The importance of methodology in astrological research.
Correlation, 1(1), 11-14.
Eysenck, H. J. (1982). The Vernon Clark experiments. Astro-Psychological Problems,
1(1), 27-29.
Eysenck, H. J. (1982). Methodology in astrological research. Astrological Journal, 24(2),
76-83.
Eysenck, H. J. (1982). Critical commentary on the Mars effect. Zetetic Scholar, 9, 61-3.
Eysenck, H. J. (1982). A note on Michel Gauquelin's "Suggestions for studying ordinary
people." Correlation, 2(2), 2-3.
Eysenck, H. J. (1983). Response to Kurtz's review of "Astrology: Science or
Superstition?" Skeptical Inquirer, 8(1), 89-90.
542 Further Eysenckian interests

Eysenck, H. J. (1983). "You, the Jury:" Are astrologers also prejudiced? Astro-
Psychological Problems, 1(2), 4—9.
Eysenck, H. J. (1983). [Astrology and] Happiness in marriage. Astro-Psychological
Problems, 1(2), 14-24; and 1(3), 15-19.
Eysenck, H. J. (1983). Methodological errors by critics of astrological claims. Astro-
Psychological Problems, 1(4), 14-17.
Eysenck, H. J. (1984). The Mars effect and its evaluation. Astro-Psychological Problems,
2(2), 22-27.
Eysenck, H. J. (1985). Parapsychology. In A. Kuper & J. Kuper (Eds.), The social
science encyclopedia. London: Routledge and Kegan Paul.
Eysenck, H. J. (1986). Critique of [Carlson's] "A double-blind test of astrology." Astro-
Psychological Problems, 4(1), 7-8.
Eysenck, H. J. (1986). James Randi and objective research. Astro-Psychological
Problems, 4(3), 7-11.
Eysenck, H. J. (1987). Aims of CORA. Astro-Psychological Problems, 5(1), 4.
Eysenck, H. J. (1987). Value of research in astrology. Astro-Psychological Problems,
5(1), 6-8.
Eysenck, H. J. (1988). Letters on objective refereeing and on statistical analysis. Astro-
Psychological Problems, 6(2), 32-33.
Eysenck, H. J. (1991). Michel Gauquelin (obituary). The Independent (London) 20 June
1991, p. 31.
Eysenck, H. J. (1991). Review of West's "The Case for Astrology." Personality and
Individual Differences, 12, 1122.
Eysenck, H. J. (1996). Sceptical prejudice [Letter]. The Psychologist, 9, 537.
Eysenck, H. J. & Nias, D. K. B. (1988). Statistical pitfalls [excerpt from Astrology
Science or Superstition?]. Astro-Psychological Problems, 6(1), 4-5.
Reports of the Eysenck astrological research seminars appear in Correlation, 1986,6(2),
5-6; 1987, 7(1), 2; 1988,8(1), 3-^; and in Astro-Psychological Problems, 1986,4(3), 4-
6; 1987, 5(2), 4-5 with pictures; 1988, 6(3), 9-17 including panel discussion.
Chapter 24
Eysenck as teacher and mentor

A. R. Jensen

1. THE CONCEPT OF INFLUENCE

One measure of a scientist is influence. It is a common standard of comparison

in many fields—in politics, the arts, philosophy, scholarship, and science. For
scientists, in particular, an objective index of influence is citation by other
scientists in scholarly journals and books (not necessarily publicity in the mass
media). Besides the sheer number of citations, there is also their average "half-
life," that is, for how many years (or decades) after their publication are a
scientist's papers and books still cited? A flurry of citations of a work can be
merely a "flash in the pan," dwindling soon to zero, while some items maintain
an imperceptibly declining citation rate for decades; few become a landmark
classic, usually mentioned without a reference.
According to the Social Science Citation Index (SSCI, which began publi-
cation in 1970), the most frequently cited among living persons in 1997 is Hans
Eysenck. Among all persons, living or dead, ever listed in the SSCI, Eysenck's
citations are exceeded only by Freud and Marx. Hence this standard criterion,
at least, stops any argument over Eysenck's influence in contemporary
psychology and the social sciences. It is a mistake to suppose that the influence
indicated by citations can be attributed simply to having published an
extraordinary number of books and articles. (The number of Eysenck's
publications, incidentally, exceeds that of any other figure in the history of
psychology.) Although there is, in fact, a high correlation between people's
number of publications and their number of citations, the correlation is less
than perfect. And, in any case, it is not directly causal, any more than the total
number of a composer's compositions determines the frequency with which his
works are performed. Siegfried Wagner, for example, wrote more operas than
did his father, Richard; and Antonio Salieri, in his 75 years, composed more
music than his contemporary, Mozart, composed in his 35 years. An example in
psychology is America's most prolific psychologist in research and publications,
Edward L. Thorndike. He died half a century ago, but in recent years is still
among 100 most often cited in the SSCI. (The same is true for Charles
544 Further Eysenckian interests

Spearman.) One of Thorndike's contemporaries with nearly as many

publications is not even listed in the SSCI. (Self-citations, of course, don't
count.) Clearly, a given work, however regarded, must make an authentic dent,
fill a gap, or initiate something new in its field to earn more than a few
citations.
Biographical facts and psychological theories about the causal factors
apparently involved in exceptional influence now have a surprisingly extensive
literature; much of it has been referenced and expertly reviewed in Dean
Simonton's (1988, 1994) two most fascinating books. (Simonton is himself the
most prolific contributor to this literature.) One could appeal to such material
for an explanation, albeit only in general terms, of Eysenck, or anyone who has
achieved eminence. But to winnow all of these generalities about the origins of
eminence to discover which ones best fit a particular subject, the specifics of
the subject's life must be sought in biographical material. The specifics on
Eysenck are provided in Gibson's (1981) frank biography and in Eysenck's
autobiography (1990). It is a rare biography or autobiography, however
factually accurate, that makes its readers feel that the person more than the
persona has been revealed, and in that respect these two works do not seem to
me exceptional. But their perspectives nicely complement each other and both
should be of intrinsic interest and value to anyone in the behavioral sciences.
Taken together, they convey an overall picture that well matches the
impression of the man and his philosophy of psychological research that
those who have observed Eysenck at close range for a number of years have
come to know.
As this kind of information about Eysenck is already accessible, therefore,
my assignment here can best supplement the extant biographical material only
if I take a more personal and subjective slant and try to explain Eysenck's
special influence as a teacher and mentor: a subject not really treated either in
Gibson's biography or Eysenck's autobiography. I imagine there are probably
many other students who have gone through Eysenck's department with an
experience somewhat similar to mine. So, with the reader's forbearance, my
story here becomes unavoidably autobiographical.
I began thinking and recollecting about Eysenck's influence when, a few
years ago, I received a letter from a historian of psychology inquiring if there
was anyone in my life I considered as personally having had an important
influence on my own career in psychology. The one and only name that
instantly came to mind was "Eysenck." I had to think a little to consider other
possibly influential persons. And there were others, of course, such as my
major professor under whom I studied for my Ph.D., Percival Symonds, and for
whom I was a graduate research assistant for three years. I have always felt
grateful for the rather fatherly and tutorial interest he took in me as a student.
A learned scholar and prolific writer, Symonds was an excellent model of
professorial and productive work habits. But then, when I began to think about
 Eysenck as teacher and mentor 545

all of my teachers and professors in terms of how different my career might

have been had I never known them, Eysenck clearly came out far ahead.
Taking a postdoctoral fellowship in his department surely seemed the smartest
thing I ever did as far as its influence on my subsequent work was concerned.

2. DISCOVERING EYSENCK

It was Professor Symonds who first suggested that I apply for a National
Institute of Mental Health postdoctoral fellowship. Take it somewhere, he
advised, that I could learn a lot more than I already knew about whatever line
of psychological research interested me the most. Symonds had recommended
me to his old and admired friend Gardner Murphy, then Director of Research
at the famous Menninger Clinic. (I wonder what course my subsequent career
might have taken had I followed that possibility.) But during my internship in
clinical psychology, I came to the happy realization that working with people
seemed much less interesting to me than working with psychological data. I
doubted that I could have been happy, or even successful, as a clinician; the
idea of becoming a professor and researcher seemed just the right ticket.
Now I must backtrack a few years, to an occasion in 1951, when I first heard
the name "Eysenck," since the circumstances probably had some valence in
determining later events. While teaching high school biology in San Diego, I
was also working for a master's degree in psychology at the State University
there. The psychology department sponsored a lecture series, which I regularly
attended one evening each month, to hear an invited speaker talk about his or
her own research. For some benevolent (and seemingly prophetic) reason that
was unknown to me, one of my professors invited me to join him and two of his
colleagues at a restaurant where they and the guest speaker (and his wife)
would have dinner before the evening lecture. The guest that night happened
to be Roger Russell, who, though an American, had recently been appointed as
successor to Sir Cyril Burt, as Professor (and Head) of the Psychology
Department in University College, London—obviously an outstanding
achievement, as Russell was then not yet 40. All I can recall of Russell's
dinner conversation were his responses to questions about Burt and then about
Eysenck. Burt was famous, of course, but this was the first time I heard of
Eysenck. (At that time, he was 35 years of age.) One of the professors at the
dinner had read Eysenck's first book, Dimensions of Personality (1947) and was
curious to know what Eysenck was like in person. Russell explained that he
seldom saw Eysenck, because he had his own department in the Institute of
Psychiatry, some miles away on the other side of London. But he went on
talking about Eysenck for a minute or two, saying he thought of Eysenck as an
exceptionally up-and-coming young fellow with the kind of ability, ambition,
546 Further Eysenctdan interests

and confidence to have already earned quite a reputation in Britain's

psychological circles. He was somebody to keep an eye on as most apt to
make a mark. And that was it.
Very soon thereafter, as a graduate student working for a Ph.D. at Columbia
University, I read Eysenck's Dimensions of Personality', then I came across his
popular and provocative Uses and Abuses of Psychology (1953). I enjoyed both
books immensely, especially for their didactic clarity and their straightforward
and logical arguments. Not in their specific content, but in their tough-minded,
no-nonsense polemical style, Eysenck's books reminded me of the first
psychology book I ever read, by John B. Watson, which I had enjoyed reading
when I was in high school. It was the origin of my interest in psychology. When
I mentioned my interest in Eysenck to Professor Symonds, he handed me a
review copy he had recently received of Eysenck's The Scientific Study of
Personality (1952). In an early chapter, one of Symonds' own studies (applying
psychoanalytic interpretations to prpjective techniques) was, to put it mildly,
trenchantly criticized. Because, as Symonds' research assistant, I was engaged
in this very kind of work, Symonds asked me to read the book and let him know
what I thought. I found this book the most exciting of anything I had yet read
by Eysenck, especially for its vision of how personality research could be
approached with the objective, quantitative, and experimental methods of the
natural sciences. I quite enjoyed Eysenck's attacks on the kind of things being
taught in some psychology courses that I had already begun to dislike as falling
beyond the pale of science. When I came back to Symonds, a little
apprehensively, to tell him my favorable opinion of Eysenck's book, I was
surprised by his nonchalant, nonargumentative response; rather he seemed
somewhat amused by my enthusiasm, and only complained about what he
thought was, in his words, "Eysenck's loudmouth style of criticism." My Ph.D.
dissertation, aimed at empirically testing some of Symonds' "dynamic" inter-
pretations of aggression in the Thematic Apperception Test, failed to
substantiate a single one of the objectively testable hypotheses derived from
Symonds' type of psychoanalytic theory. But this didn't seem to disturb him; he
remarked, good naturedly, "You seem to have overworked the null
hypothesis." (My sample sizes were sufficiently large and the statistical results
were sufficiently clear-cut to counter Symonds' question about Type II error,
and all of my thesis examiners, who were reputedly tough—Professors Irving
Lorge, Edward J. Shoben, and Joseph Zubin—seemed to like the findings.
Symonds was supportive throughout, and even suggested that I submitted a
shortened version of my thesis to Psychological Monographs, which I did. Later,
he offered to write a recommendation to NIMH, knowing I intended to apply
for a postdoc in Eysenck's department. But he advised that I should ask people
who might have some inside knowledge of what I might expect of Eysenck. I
think there is perhaps a general lesson to be learned from my having followed
that suggestion. I inquired of three professors whose opinions I respected.
 Eysenck as teacher and mentor 547

Though they all knew something about Eysenck's work, only one had ever met
him in person, and none had any inside knowledge of Eysenck's department or
knew how postdocs fared there. And it turned out that the suppositions
proffered by two of them were way off the mark.

3. SOME OPINIONS, CONJECTURES, AND MISCONCEPTIONS ABOUT EYSENCK

3.1 Another Thorndike?

Professor Laurence Shaffer, who headed the clinical psychology program at
Columbia and, in his capacity as president of the APA, had met Eysenck on his
1953 visit to the U.S.A., was by far the most enthusiastic. He said he thought
Eysenck was much like E. L. Thorndike when he was Eysenck's age, and that
this presaged comparable eminence. Neither E. L. Thorndike nor Eysenck, he
said, was an eclectic who tries to find a little good in every viewpoint, and they
were similar in promoting their strong convictions about the path psychology
should follow, disdaining anything that was contrary. I was happy to hear
Shaffer's likening of Eysenck to E. L. Thorndike, because at that time
Thorndike (probably still America's greatest psychologist) happened to be my
only "hero" in psychology. (He still is, but there are now also a few others.)
Shaffer admitted that he himself wished he were in a position that allowed him
to spend a year in Eysenck's department; first choice, absolutely, he said, and
he offered to recommend me.

3.2 Another Burt?

Professor Robert L. Thorndike, on the other hand, warned that Eysenck might
hold such strong opinions about psychology as to be too authoritarian and
intolerant of anyone who didn't completely agree with him. He said he had
once heard his father, E. L. Thorndike, speak about Cyril Burt to the effect
that he was notoriously authoritarian; and, after all, wasn't Eysenck Hurt's
protege? And his leading disciple? (How little Thorndike knew of Eysenck's
rebellious relationship to Burt!) Not long after I arrived in London, I learned
that this authoritarian image of Eysenck could hardly have been more false. In
fact, I was surprised by Eysenck's attitude toward his postdocs (of which I was
one of several during my two-year stay) and even his own staff. It appeared all
so nondirective and laissez-faire. Yet nearly everyone seemed to be working on
things related to Eysenck's research program. Naturally, I thought; and why
shouldn't they? I got the impression, however, that Eysenck would be little
interested in anyone who wasn't engaged in research at least consonant with
his view of psychology as a quantitative natural science. Yet I always felt I
could believe, say, and do whatever I pleased during my two years' postdoc
without anyone's even noticing or asking. Eysenck's attitude toward personal
548 Further Eysenckian interests

one-on-one interaction with postdocs was entirely passive. One had to go to

him, not the reverse. When my research was related to his interests, he was
immensely helpful, providing laboratory equipment, experimental subjects,
and any advice I ever asked for. It was as near an ideal learning environment as
I could imagine.

3.3 Aloof and remote?

Professor Joseph Zubin had surmised, judging from Eysenck's extraordinary
output, that anyone who kept himself that obviously busy would most likely be
stingy with his time and attention to others, particularly postdoctoral fellows.
Zubin imagined Eysenck's department was probably very hierarchical (as was
traditional in many European universities), so that outsiders like me, assigned
to the bottom rung, would hardly have access to "the great man" at the top.
And Zubin gave me a copy of a highly detailed "inside-report" of the U.S.
Office of Naval Research (which had awarded Eysenck a large grant) about
what was going on in Eysenck's laboratory. The report had been prepared by
Professor Lee J. Cronbach during his tenure (1955-1956) as the ONR Liaison
Officer in London, and I found it a fascinating document that gave more
informal inside information about Eysenck and other members of the research
staff than I had been able to find elsewhere. I already knew Cronbach's
reputation as an exceptionally sharp but rather ungenerous and acerbic critic
who pulled no punches. Yet I found his ONR report quite favorable over all,
and it left the impression that Eysenck ran a lively shop, very much as I had
gathered from reading his books and journal articles. Hence I felt more certain
that I should head for London. When finally I got there, I discovered first-hand
that Professor Zubin's reservation about Eysenck as possibly remote and aloof
(at least to postdocs) was quite the opposite of the prevailing conditions. (In
talking with postdocs at other institutions, however, it was common to find
their initial hopes disappointed by their meager opportunities for interaction
with their nominal mentors.)
I found Eysenck to be the most dependably accessible professor in all of my
experience, both before and since my postdoc. From 8:30 am to 4:00 or 5:00
pm, Monday through Friday, Eysenck was in his office. His door displayed one
of three signs: either "In," or "Dictating," or "Out." The "In" sign was most
often displayed in the afternoon; then, if you wanted to talk with Eysenck, you
only had to knock on the door. Glad to discuss any technical question or
problem, he did so with admirable clarity and authority, thoroughly but
efficiently, plainly avoiding any time-wasting pleasantries or small-talk—it was
all pure business. (He never in the least took what could be called a personal
interest in anyone around the department, as far as I could tell; in fact, his
totally impersonal attitude toward everyone so irritated one secretary that she
quit her job because of it!) Observing Eysenck's whole routine, I always got the
 Eysenck as teacher and mentor 549

impression he was one who, as musicians say, never missed a beat. But he was
never hurried, always relaxed, seemingly easy-going, and, strangely, he never
even seemed at all busy when one entered his office (for example, I rarely saw
anything on his desk, and his phone calls were controlled from the secretarial
office). Yet, every day incredible amounts of work emanated from his office,
keeping two full-time secretaries constantly busy. The smooth, frictionless
efficiency of Eysenck's operation was indeed impressive.

3.4 Ferocious?
Dr Michael Shepherd, a psychiatrist from the Maudsley Hospital, one day
came as a guest lecturer to the University of Maryland Psychiatric Institute in
Baltimore, where I was on my clinical internship. So I asked him, too, about
Eysenck—what was he like? "Oh, a brilliant and charming fellow," Shepherd
said, "Ferocious only in the face of opposition!" Ferocious1} It could make one
think twice about going to his department. But what a terribly wrong
impression this was—as I later discovered! The notion of "ferocious" was a
misnomer, or at best a kind of half-truth, and quite exaggerated at that. It was
wholly misleading in any personal sense, but I discovered it had become a
popular image outside Eysenck's department, and especially in British
psychiatric circles. It certainly didn't apply to any of Eysenck's personal
encounters, with me or anyone I knew who ever had any personal dealings with
him. This fictitious reputation seems to have originated, however, from two
indisputably real sources: First, there was Eysenck's hard-hitting, but right-on-
the-mark, published criticisms of certain views in psychology that, from his
perspective, looked more like religion than science. And second, there was
Eysenck's supreme confidence and his quickness of mind and mastery of his
subject that he could summon in a flash to deftly impale anyone who would
stand up at the end of one of his lectures and dare to denounce or argue some
point. In almost every audience at one of Eysenck's open lectures, it seemed,
there was someone who wanted to oppose him on one point or another and
imagined he could nail Eysenck with a clever put-down type of question or
argument. (One inveterate opponent even advertised widely his intention to do
this.) Eysenck's lectures often seemed to invite these attacks. Evidently most of
his critics had no warning or premonition of the likely outcome, although the
typical outcome had become such common knowledge as to be almost
legendary. Smiling as if he relished the moment, Eysenck invariably answered
his critics in a conspicuously courteous manner—serene, and in the same
modulated voice as his lecture. But the amazingly pointed, cogently organized,
and perfectly articulated assembly of facts, logic, and argumentation that
Eysenck could immediately level against his critic's statement was generally
viewed by the audience as a knockout blow. Usually that ended the encounter,
with the derailed challenger taking his seat, looking rather put out, and saying
550 Further Eysenckian interests

no more. Whether or not the poor fellow (it was always a male) felt humiliated,
the audience perceived it as such. And the most tender-minded among them
probably interpreted Eysenck's broadside as "ferocious," however perfectly
cool and seemingly polite. I heard someone once joke that it all went off so
neatly that he wondered if a stooge was planted among the audience, or if
precognition permitted Eysenck to have prepared and rehearsed his response
some days in advance. This slyly combative facet, however, was so unlike
anything I ever noticed in his encounters with students, or his seminars, or his
"at homes," or in private conversation, that I regarded it as a kind of
showmanship (along with a good bit of what Stephen Potter referred to as
lifemanship) reserved for special kinds of opponents, and then deployed only
when there was a large audience present. In his later years, he has either toned
it down noticeably or his more aggressive opponents have dwindled—probably
both. The last quintessential knockout I witnessed was at one of his lectures in
Australia, in 1977, when we were both invited there for a series of
presentations at several universities. Eysenck always seems so laid-back at
these events that I suppose he scarcely remembers them, though they become
other people's anecdotes.

4. DISTILLING MY POSTDOCTORAL EXPERIENCE IN EYSENCK'S LAB

4.1 Introduction
So that's the story of how my two-years' postdoc with Eysenck (1956-1958) all
came about, and, of course, it was only made possible by Eysenck's kind
consent and a generous fellowship awarded by the NIMH. I had found the
whole experience so rewarding that six years later, on my first sabbatical leave
(1964-1965) from my chair at Berkeley, I applied for and received a
Guggenheim Fellowship to revisit Eysenck's laboratory for the whole year.
So now, from my store of memories of the three years, in all, that I spent with
Eysenck, I shall try to distill out those elements that I think contributed most to
his profound influence as a teacher and mentor.

4.2 Eysenck as a writer

Many people have been influenced by Eysenck only in his role as a writer. And
probably even among the circle of psychologists who know him personally, his
main source of influence is his books and articles—an influence that reaches a
number of different audiences, since his research and writing have made
significant contributions in four major fields in the behavioral sciences
(personality, behavior therapy, human abilities, behavior genetics), as witness
the variety of topics covered in the present volume and in the collection of "pro
and con" essays on Eysenck's work edited by Professors Sohan Modgil and
Celia Modgil (1986).
 Eysenck as teacher and mentor 551

Besides the purely factual and theoretical contents of Eysenck's output,

which, of course, are the main interest, the one aspect that I think hooked me
and has probably had the most generalized and continuing effect is a
characteristic manifested in nearly all of his writings, particularly his books and
his many contributed chapters (even his reviews of other authors' books). I
refer to the fact that they are infused with a philosophy of science, usually
illustrated with analogies between research problems in psychology and those
in other sciences. The subtly didactic quality of his writings results mainly from
his typically going a step beyond any specific finding or fact to a more general
level regarding method and theory, so the reader comes away with a rather
more generalizable order of knowledge and understanding. His writings are
also imbued with certain attitudes, values, ideals, and inspiration that more or
less unconsciously carry over to one's own work. This is the quintessence of
intellectual influence.
It helps, too, that Eysenck's writing style is so clear and easy, especially when
he has to explain complicated things. He himself refers to it as a "natural"
style, probably because he writes exactly as he speaks (or vice versa); I've never
known anyone else whose manner of speaking and writing are so much the
same.
His speaking and writing are, in fact, one and the same, as I discovered
during my postdoc. Eysenck's steady output of publications is so incredible
(almost three times that of E. L. Thorndike, who held the world's record for
number of publications in psychology until Eysenck came along) that, as a
postdoc, I was curious about how it was possible. It so happened that Eysenck's
office was clearly visible, at a distance of about 50 feet, from the window of my
office, at an angle of about 135 degrees. The large glass French doors that
opened onto the balcony extending from Eysenck's office afforded a full view
inside. I noticed that nearly every morning, for about three hours, he paced
around in circles in his large office (he seldom sat at his desk), dictating the
whole time, either to a secretary taking shorthand or to a dictaphone on his
desk. Two secretaries were kept busy all afternoon typing whatever he dictated
in the morning. During the first half-year of my postdoc, Eysenck dictated two
books and many journal articles and book reviews, in addition to many
research proposals, progress reports, and a large correspondence. One day his
secretary couldn't make out a technical term in his recorded dictation and
asked me to put on the earphones and listen to the playback at that point.
Eysenck's dictation was paced just as if he were simply reading aloud,
smoothly, evenly, without any back-tracking, double takes, corrections, or
hesitations. And the resulting typescript, except for correction of a few typos,
was sent off to the publishers. When I once mentioned to Eysenck his amazing
skill at dictation, he said it was simply a gift for which he was most grateful.
Someone suggested that it resulted from the fact that English was Eysenck's
third language (after German and French), which he had acquired mainly from
552 Further Eysenckian interests

reading psychology books in English and therefore didn't have to suppress

informal and colloquial language as most of us do when we write (or try to
dictate) formal prose. One summer, after Eysenck had dictated a whole book
during two weeks of his summer vacation on the Isle of Wight, I asked him if
he ever got tired while dictating for hours at a stretch. A moment's thought,
and he pointed to his jaw, saying that perhaps at times his jaw got a little tired.
The jaw, not the brain! Pure Eysenck! His secretary played me a part of one of
the dictaphone tapes he had brought back from his vacation, again hoping I
could help decipher one sentence that was obscured by—what?—voices on a
TV show that one could hear in the background! Apparently the TV didn't in
the least hinder his concentration or the easy flow of his words forming well-
constructed sentences and paragraphs! It all could be published without
editing. He usually left mere proofreading entirely to others. Occasionally this
resulted in a minor catastrophe; for example, a proofreader rather consistently
altered the spelling of Spearman's term noegenesis, making it neogenesis
instead, in a work (Eysenck, 1979) that, overall, is probably still the best college
textbook on intelligence. (In light of the boom in research on human mental
ability since 1979, a new, updated edition of Eysenck's 1979 book would be
most welcome!)
What might have been an example of Eysenck's productive efficiency
occurred at one of his office seminars, attended by his five postdocs and any
members of his staff who wished to attend. He began the discussion by
delivering, off-the-cuff (as no notes were in sight), what amounted to a full
lecture on the history of personality research. I thought it all so beautifully
organized and clearly delivered it seemed a pity that it wasn't tape-recorded,
typed, and submitted for publication. But then apparently that is what he had
done, either before, after, or possibly during, his lecture (I didn't notice
whether his recording machine was "on"), because several months later it
appeared as an article in the British Journal of Psychology. As best I could
recall, it was word-for-word the same as the spoken presentation.
Besides their substantive content that makes Eysenck's writings important,
an added attraction is the absence of the jargon and the dry, bland, and
uncommitted tone that prevail in academic psychological literature. For those
writers afraid of criticism, a studiously impersonal, pedantic, tentative, and
timid style might possibly help in warding off intellectual opponents. Eysenck,
on the contrary, likes to stick his neck out; excessive caution for fear of
criticism is not one of his faults. Avoiding the neutralizing stylistic devices often
adopted in academic writing to mute or obscure potentially controversial
statements, Eysenck's writing (and speaking) style, though never in the least
dogmatic or doctrinaire, maintains an explicit viewpoint on the topic at hand,
states clear-cut opinions, shuns hedging, displays clarity of factual exposition
and cogency of argument, and delivers a strong message. Many readers admire
this. But I have also come across a few persons who are riled by it; and if they
 Eysenck as teacher and mentor 553

also happen to disagree with the substance of the message, they become
furious. Eysenck commented once, with amused wonderment, that for some
people, for some odd reason, his writing apparently has "emotional stimulus
value." I recall one such incident, at a luncheon in Berkeley with a group of
psychologists at the Institute of Personality Assessment and Research (IPAR).
We were entertaining a guest speaker, a British psychiatrist of Freudian stripe,
who was scheduled to give a lecture at IPAR that afternoon. He had every
appearance of a dignified, intellectual, and amiable gentleman—until, at one
point during the lunch conversation, someone happened to mention Eysenck's
then current book, The Dynamics of Anxiety and Hysteria, whereupon our guest
flew into an emotional tirade, face reddening, eyes bulging, as he sputtered
denunciations of Eysenck's book and its basic conception of psychology as a
quantitative, experimental science. With everyone looking anxious, as if fearing
the speaker's emotion could escalate into an apoplectic stroke, the luncheon
hosts, Professors MacKinnon and Crutchfield, abruptly intruded and changed
the topic completely, while our rattled guest regained composure. Such is
Eysenck's peculiar effect on what one hopes are a rare few. (The only person in
my experience who had even stronger negative "emotional stimulus value" for
some people was Professor William Shockley, but that's another story.)

4.3 Eysenck as teacher

Although Eysenck never seemed busy, one knew from all the things that were
happening in his department that he really couldn't have been other than busy,
and therefore one always felt a little reluctant to impinge on the time he spent
in his office. I would go to him only when I wanted his own opinion in
particular on some point. The Institute had other expert research psychologists
and two excellent statisticians (A. E. Maxwell and Patrick Slater) to whom one
could go for help and discussion of problems. When necessary, one could go to
Eysenck as a final arbiter; and it was always clarifying and enlightening. I
would come away feeling I had witnessed a formidable intellect brought to bear
on the given problem. It was something like a student of conducting (as I once
was) watching Toscanini in rehearsal with the NBC Symphony (as I did many
times while I was a student in New York.)
Eysenck also received an extraordinary number of visitors. Each took some
of his time. It seemed (especially in summer) that every noted American
psychologist who came through London wanted to see Eysenck. The list of his
visitors was like a "who's who in psychology." He would spend an hour or so
talking with a visitor; through my office window I would see him standing at the
blackboard in his office, drawing graphs and explaining things to his visitor.
Occasionally the visitor was invited to give a research colloquium for the staff
and postdocs, or we would be able to meet a visitor in informal discussion at
554 Further Eysenckian interests

Eysenck's weekly "at homes." In my two postdoctoral years there, I heard and
met more famous psychologists than I ever saw at any one APA convention I
have ever attended. What a place to take a postdoc!
The graduate students studying for a Ph.D. in Eysenck's department were
assigned to work under the supervision of one or another member of the
research staff; I can't recall any evidence of Eysenck's personally paying them
much attention, other than his initially going over the student's research
proposal, then reading the dissertation and attending the student's oral
defense of it. Eysenck dispatched such chores faithfully; he never allowed his
work to pile up. In two instances I witnessed, drafts of dissertations delivered
to him by students on one day were returned on the very next day,
accompanied by his remarkably detailed corrections, queries, and comments.
Most fortunately for his postdocs, however, Eysenck allowed many informal
opportunities for us to talk with him and to sit in on his discussions with others.
The daily routine was for everyone to go over to the Maudsley cafeteria for a
half-hour's mid-morning coffee break. In order to discuss something with one
of his research staff without having to infringe on their office time, Eysenck (if
not giving dictation) would accompany the person to the cafeteria, and as many
of us as could crowd around the same table would be able to get in on the
discussion. This occurred, on average, two or three days a week. Eysenck never
took coffee or anything else (he explained that he hated the cafeteria coffee,
food, etc.); but for him this wasn't a coffee break, but a concession to the
efficiency of his department's research mission. These sessions were like brief
research conferences, in which one could learn how Eysenck thought about
many things, though always psychological. In the aggregate, over two years, I
found these sessions were more revealing of Eysenck's character than any
other form of his encounters with people that I was ever in a position to
witness. (I was struck, incidentally, by how much politics was discussed at other
tables during coffee breaks, but never around Eysenck; in his presence, no one
strayed from psychological research or scientific matters; he himself seemed to
have no other intellectual interests.)
Eysenck's only formal teaching duty, to my knowledge, was the series of
lectures he gave as part of the Institute's training program for psychiatric
registrars. I attended these, and although they covered rather elementary
psychology (and much of the material in Uses and Abuses of Psychology), I
found it interesting to see how neatly Eysenck presented it. Another postdoc,
who also attended a few of these lectures, complained that they went off so
effortlessly and automatically (though Eysenck never read from notes) as to
give the impression that he wasn't really thinking out his lecture then and
there; and the registrars' few simple questions usually permitted such easy,
one-sentence answers as to be uninteresting. True, a psychology postdoc could
hardly get much from these formal lectures that the Institute required of
 Eysenck as teacher and mentor 555

Eysenck. They were designed for an entirely different kind of audience. In a

sense, there is a fundamental difference between formal lecturing and informal
teaching, and this distinction seemed especially clear in Eysenck's case.
His less formal teaching function, intended specifically for his postdocs, was
entirely voluntary on his part. It would take place in his office seminars on late
Friday afternoons, usually attended also by certain members of his staff,
depending on the topic of discussion, which was most often related to their
current work in Eysenck's research program. In this setting, Eysenck rarely
lectured but was truly a teacher, doing the things that advanced students can
most profit from—providing a mix of ideas, hypotheses, constructive criticism,
suggestions, discussion, argument, questions, and answers on the topic of that
day's seminar. Each session lasted over two hours. Afterwards we would
usually go to the local pub for a pint of bitter, without Eysenck, of course, as he
was very abstemious (once saying that he avoided anything that depressed
cortical activity); on several occasions I've heard him say that he particularly
hates beer. I doubt that he's ever been inside a pub.
Eysenck's most distinctive characteristic as a teacher, as in much of his
writing, is that he always uses the particulars of a given problem to point out
certain general rules that they suggest. Besides discussing the specifics of a
given problem or technical point, he habitually elevated the specifics to a
conceptually higher, more general level than they possessed when first
introduced. Hence, always along with any factual information, one imbibed
general concepts and principles, which over time coalesced finally into a whole
philosophy for behavioral science research, embracing taxonomy, objective
measurement, statistical reasoning, a reductionist orientation with tie-ins to
genetics and biology, hypothetico-deductive methodology, and a strong
conviction—a moral ideal—that scientific inquiry, properly worked as a
perpetually self-correcting process, will allow natural truths to get out. In the
history of science even some of the most improbable and zany hypotheses,
properly worked out, have led to important discoveries. This mental attitude, I
think, is the essence of Eysenck's personal influence on one's subsequent
career. It is something to live by. Few professors I have known, even if they
wished, would be able to imbue students with this sort of ideal in such full
measure as Eysenck was able to impart. This is because it was never delivered
all at once on any occasion, like an explicit lesson or sermon; rather it was the
basic operating principle that infused Eysenck's own thinking and activity; so it
was inevitably and thoroughly, yet almost imperceptibly, disseminated
throughout the many occasions spent with him. Such is the influence of an
ideal mentor.
556 Further Eysenckian interests

4.4 Eysenck's "at homes"

Eysenck's life clearly revolves around his work, and many of us in his
department were the beneficiaries. A special treat that I always looked forward
to were Eysenck's "at homes," held regularly every Wednesday night, from
7:30 to 10:30. The research staff and postdocs had a standing invitation. These
meetings were most informal, with a good deal of banter and joking, and Sybil
Eysenck always served refreshments. But the "at homes" were also highly and
pleasantly instructive. Typically, after everyone had comfortably settled down
in the living room, someone was prepared to describe, in a quarter of an hour
or so, the theoretical rationale and the methodology or procedure (and the
results, if any) for the study or experiment he or she was planning, or was
already engaged in, or had just completed. Others would chime in with
questions and comments, and Eysenck would often bring up any closely related
studies or theories in the literature. (He has an amazingly encyclopedic
knowledge such as one rarely encounters in a lifetime.) Alternative methods of
statistical analysis would be considered, and after an hour or so the subject had
received a fairly thorough discussion. Members of the group also would
volunteer comments on recent journal articles or books relevant to our
research interests. Eysenck covered anything of interest in the German
psychological journals, which he routinely read, and nearly every week he
could report on some recent book he had just read. He is a voracious reader;
few, if any, psychologists I have known are in the habit of reading even half as
many books as Eysenck routinely publishes reviews of. It's a rare month that I
don't find at least one book review by Eysenck in some journal. Often the
review is more worth reading than the book itself. I have yet to come across
another reviewer in psychology, except Cyril Burt, who brings such a broad
erudition to bear. Besides informing about the book's contents, of course,
Eysenck's reviews are also used as vehicles for exercising in various ways his
own philosophy of psychology. (What other good reason could there be for the
thankless chore of writing a book review?)

5. THE BASIS OF EYSENCK'S INFLUENCE

Individuals differ greatly on the scale of innate capacities, and if measured on

the same scale, individuals' achievements differ even by orders of magnitude.
Gallon attributed the level of outstanding achievement that distinguishes a
person as eminent in some field to that person's possessing a combination of
three distinct factors, each to an exceptional degree: ability (or capacity), drive
(or zeal), and perseverance (or sustained goal-directed effort in the face of
problems, hardships, or opposition). Eysenck's illustrious career manifests all
these conditions in abundance. They are the general basis of his eminence and
influence.
 Eysenck as teacher and mentor 557

As for ability, it seems vapid even to mention Eysenck's exceptional level of

intelligence, as it is so obvious. To be more specific, what I have especially
noticed (and it is a more outstanding characteristic even than his possession of
a phenomenal encyclopedic knowledge) is that he often, and quickly, grasps
the larger picture that is latent in a welter of more or less isolated facts. At
times, in his writings or in personal encounters, I would get the impression that
his mind encompasses a larger theater for performing intellectual operations
than most people's. To a greater degree than many others who are considered
highly intelligent, he discerns a larger number of the essential elements
embedded in a problem, issue, or argument, and can mentally manipulate a
larger number of these elements all at once, to arrive at a clearer, more
coherent, and larger comprehension of the issue than most of us can come up
with. It's enviable, of course, this facility to confront a novel problem and
readily see the larger picture and all its various elements in their perspective
relation to the whole. (I suppose this larger "mental space" is a part of his
ability to artfully compose an entire article or book chapter in one continuous
dictation session.) Also enviable is his ability whilst in a discussion, to so
quickly cut through and clarify any confused or muddled argument; to swiftly
scan his vast store of knowledge to bring up the most directly relevant material
on a given point; to do certain things remarkably easily that many psychology
Ph.D.s apparently never can do at all. Seeing such mental equipment, some
five sigmas out from the mean, can at first seem a bit discouraging, but then
one must learn to live with the plain fact of nature that, on any particular
dimension of human differences, some persons are given more than others,
and so you simply try to do your own best with what you have.
As for motivation or zeal, one sees in Eysenck no sign of the typical image of
a "driven" person. Quite the contrary. One of his colleagues once remarked
that if you watched him for any short period of time during the day, you'd likely
get the impression he was a quite laid-back and lazy sort of fellow. A correct
description of appearances perhaps. But such an illusion! It seems to me that
we have to infer some extraordinary, perhaps obsessive, level of intrinsic
motivation, and probably a certain inevitable egocentric selfishness in the
deployment of time and energy, to explain Eysenck's achievements and what
might be called (for want of a better word) his unflagging dedication to the
advancement of psychology as a natural science. But this energy or drive never
seems to show itself physically, except for his constantly pacing around in his
office while he is dictating. Perhaps he dissipates the purely motor aspect of his
high energy level during his daily exercise on the tennis court, which then keeps
him super-relaxed during his working hours. It is impossible to imagine the
absence of an immense and well-channeled level of energy in any person who
makes as big a dent in his field of endeavor as Eysenck has done in behavioral
science. Besides a strong personal ambition to make a mark and enhance one's
self-esteem (which is absolutely common to all great achievers), I see the
558 Further Eysencldan interests

overwhelming intrinsic motivation for Eysenck's activity as a product of

channeling an abundance of general mental energy into the typical aspirations
of a scientist to acquire existing knowledge, to discover new knowledge, and, in
general, to advance his field of science, which in Eysenck's case just happened
to be psychology. (My use of the word energy here doesn't mean ability per se,
or Spearman's g, but rather a kind of inherent lifelong itch to use one's brain,
just as some athletes speak of having an itch to use certain muscles. This "brain
itch" is an interesting variable [or construct] that hasn't yet received much
attention in differential psychology, though people differ markedly in it.)
As for perseverance, Eysenck's entire career speaks for itself. It has sustained
a constant rate of immense productivity now for over 55 years. The "greats" in
any field, as I've noticed in my reading of many biographies, typically keep on
doing "their thing" all their life, barring dire illness or infirmity. (This itself is
an interesting psychological phenomenon, since most people are glad to be
able to retire from work as early in life as they can attain sufficient financial
security, and then they retire whole hog.) Eysenck is among those elect few
that one can think of whose lifetime record of accomplishments could be
divided among a dozen persons and each of them would be considered
outstanding by ordinary standards, with various honors, biographical entries in
Who's Who and the like.
A few years ago, while having dinner with one of Eysenck's long-term
colleagues, Dr Niel O'Connor (a distinguished psychologist in his own right) he
said something that struck me as interesting and, I think, peculiarly apt.
Shaking his head in puzzlement, he said he never could think of Eysenck really
as a person, but always thought of him as some kind of institution or
phenomenon. (At these words, I remembered George Bernard Shaw's
comment about Mahatma Gandhi, spoken to news reporters; GBS said
"Gandhi is not a man, but a phenomenon.") Niel O'Connor's insight is true, I
think, at least in one sense. Of course, Eysenck is a man (as certainly was
Gandhi), but by normal standards those visible aspects of his great career are,
as they say, "larger than life"—and indeed a phenomenon.
A significant part of this phenomenon is Eysenck's influence and inspiration
as teacher and mentor. My own experience and observations related here are
but one idiosyncratic example. But many of Eysenck's former students, in every
continent of the globe, I'm sure, would like to give their testimony about how
their work in the behavioral sciences has been influenced through their own
unique experiences of Eysenck. The one thing we all would express in common
is our continuing gratitude for this privilege and good fortune.
 Eysenck as teacher and mentor 559

REFERENCES

Eysenck, H. J. (1979). The structure and measurement of intelligence. New York:

Springer.
Eysenck, H. J. (1990). Rebel with a cause: The autobiography of Hans Eysenck. London:
W.H. Allen.
Gibson, H. B. (1981). Hans Eysenck: The man and his work. London: Peter Owen.
Modgil, S. & Modgil, C. (1986). Hans Eysenck: Consensus and controversy. Philadelphia:
Palmer.
Simonton, D. K. (1988). Scientific genius. Cambridge: Cambridge University Press.
Simonton, D. K. (1994). Greatness: Who makes history and why. New York: Guilford.
This page intentionally left blank
PART IV

Epilog
This page intentionally left blank
Chapter 25
Psychology as science

H. Nyborg

1. INTRODUCTION

There is considerable confusion about how to define psychology as a science.

Some see it as a social, others as a biological science. Some universities place
psychology under the humanistic faculty, others under medical, social, or
political science. At still other universities it is part of information technology,
sociology, anthropology, philosophy, or computer sciences. Opinions also
differ about what the proper subject for psychology is. Some define it basically
from a philosophical or anthropocentric point of view, others see it as basically
a material subject. It is a truism to say that methods must be adapted to the
phenomenon under investigation, but as there is no general agreement about
the subject for psychology, the methods applied differ widely. Some use
correlational techniques, others experimental, some a nomothetic, others an
idiographic or clinical approach. Many social scientists, including cultural
anthropologists and psychologists, even doubt if there is such a 'thing' as a
human nature to be studied. Genes, biology, and brain structure and function
mean little to such scientists; they simply constitute the basically neutral
substrate on which societal forces forge the social construction of sexual iden-
tity, intelligence, personality, human nature in general, or even the recon-
struction of society.
The main title of the book—The Scientific Study of Human Nature—was
chosen with this boggling confusion in mind. So was the book cover. Here, the
fifteenth-century Dutch master Pieter Brueghel reflects on an old legend:
People who dare to construct a building that reaches into the sky approaching
God deserve punishment. What could be more devastating than forcing the
builders to speak in different tongues? They would not understand a word of
each other, and their construction would amount to nothing. Perhaps this
metaphor illustrates quite well what happened to psychology from the very
beginning. Not that God punished the builders, of course, but rather that a
self-induced punishment was enforced. These early scientists listened too much
to armchair philosophers, who with unbelievable ease produced flashing
564 Epilog

metaphors that, as a rule, never materialized into empirically testable ideas.

Their scientific construction, therefore, soon began to stumble and the design
that was necessary to support a proper science of human nature disappeared in
heavy verbal smoke. As an extra punishment, almost everybody followed rather
uncritically the advice of the early philosophers, and used them in a verbal
crusade lasting more than two-thousand years. All this developed uncontrolled
head-banging on intractable linguistic body-mind problems for those involved.
The result was the establishment of a large number of very different pseudo-
empirical schools. It was not even possible to find absolution in eclecticism,
because the schools were based on fundamentally different and often opposing
ideas about human development and nature. Fragments taken from the
various schools were bound to lose the internal construct validity of the original
movement.
This Babylonian confusion created more heat than light, and the philo-
sophical impact on the study of human nature turned into the greatest
intellectual disaster of all time. The fault was not so much that, in the absence
of proper empirical tools, the first faltering steps toward a scientific psychology
were turned into clever and rational exercises, but rather that few saw any need
to confront the clever word games with an external empirical censor to harness
them and evoke self-correcting procedures. Most of the followers accepted for
centuries that rhetoric was the measure of progress. The few stubborn
scientists that constantly resisted the temptation of easy linguistic solutions and
worked experimentally, were largely neglected. This tendency can be seen even
in contemporary psychology.
However, changes to the philosophically inspired psychological construc-
tions seem now to be induced by the recent progress in brain and molecular
sciences. In fact, psychology based on the idea of an abstract psyche begins
more and more to look like a house of cards on the brink of collapsing in the
slipstream of the very successful empirical brain sciences.
Such a bleak view on the apparently successful psychology is bound to lift an
eyebrow here and there, and the indictment is admittedly strong. The following
sections strive to justify the position, by trying to find answers to three
questions: (1) Precisely when, where and what went wrong with psychology?
(2) What is Eysenck's view of psychology as science? and, (3) What is the most
likely next step in the development of an appropriate twenty-first-century
scientific study of human nature?
Here I must ask fellow contributors to this book for permission to have a
free hand. Some undoubtedly disagree in part or in toto with my dreary view of
psychology as science. Worse, I intend to draw heavily on their more pes-
simistic conclusions and indications of obvious lacuna in contemporary
psychology to illustrate the fundamental fault in psychology in general.
 Psychology as science 565

2. WHEN, WHERE, AND WHAT WENT WRONG WITH PSYCHOLOGY AS

SCIENCE?

2.1 The fatal decision

Plato, Aristotle, and Democrit are key actors in the following simplified
account of the early formation of psychology as science. One interest they had
in common was to identify the nature of things. One of Democrit's ideas was
that there are eternally moving atoms and a vast void. Things materialize when
atoms collect. They take form in the process, and cease to exist when atoms
separate again. This materialistic position heralds a much later atom theory
and a natural-science view on the world. Actually, Democrit held additional
views that are rather unpalatable to natural science, but this need not concern
us here.
Plato and Aristotle saw the world quite differently, as can be illustrated by
the 'cave' example. Imagine a man standing in front of a fire. The shadow he
projects on the cave wall better expresses the general idea of humans and their
nature than the individual projecting it. Form is accordingly more important
than content, and the abstract is more important than the concrete. This is an
early forerunner for the later philosophical-psychological-humanistic view of
the world, where abstraction becomes the basic element in an explanation. The
metaphysics of Aristotle presumed that everything is predestined, but this and
other aspects are disregarded here.
Although they were contemporaries, Plato (427-347 BC) and Aristotle
(384-322 BC) never met personally with Democrit (460-370 BC), as far as we
know. Democrit was about 70 when Plato's pupil, Aristotle, was a boy of 14,
but they probably knew each others' positions quite well, as their time was one
of fierce discussion of these matters in several ancient places.
Unfortunately, Democrit and his materialistic view by and large lost the
battle for the scientific study of human nature. Gradually, it became generally
accepted that concept formation, logic, and rationality were the proper tools
for getting on with meaning, abstraction, theory, and philosophy in order to
define (human) nature.

2.2 The dire consequences

The early defeat of the Democritean view may, to a large extent, be seen as the
basic problem in contemporary psychology. It was all very well that, after a
while, confused ideas of an animated nature gave way to a God-given soul that
was later renamed psyche (or ego, self, or me), to rid it from religious
connotations. It was also taken as progress that psychologists then renamed the
psyche cognition or metacognition, and that other scientists invented highly
abstract superorganismic concepts like social norms, cultural stereotypes, and
even collective consciousness. In reified form these concepts were gradually
 566 Epihg

acknowledged to either represent or exert an indirect causal impact on mind

and human nature.
The common theme behind these apparently very different phenomena is
that they are based on notions of abstract Platonic qualities, which elevate
them to a status above the material brain or world. The defeat of the
materialistic view of Democrit thus paved the way for the extensive use of high-
level abstractions as a substitute for proper operationalizing and explanation in
the scientific process. Inevitably, this easy solution led to excessive verbosity in
the form of uncensored fabrication of hypothetical constructs and intervening
variables, reification of these variables, the postulation of causal relations
among them, and to the various forms of mentalism-based philosophies that
characterize most forms of social sciences, including psychology.
This briefly characterizes (some may perhaps say, makes a caricature of) the
actors and sets the stage for the creation of the numerous insurmountable
philosophical-psychological dilemmas associated with a dualist view. The
results of this conceptual dance macabre with human nature as the victim, is
illustrated graphically in Figure 25.1.
According to Figure 25.1 we have in principle four different types of analyses
at our disposal after the great intellectual disaster: Surface, top-down, bottom-
up, and all-bottom approaches. The three in the middle are the children of the
body-mind schism and demand hierarchical solutions. Surface analyses are
basically of a purely descriptive value, usually at a very high level of abstraction.
They may, nevertheless, serve as a useful starting point for the formulation of
genuine causal questions. All-bottom analyses circumvent the dualist traps and
allow for proper causal analysis.

Approach

All surface analysis

Gene-environment
analysis

All bottom analysis

Figure 25.1. Types of analytic approaches to abilities and personality (from Nyborg, 1995).
 Psychology as science 567

Examples of surface approaches are behaviorism, information theory, social

learning theory, phenomenology, cultural anthropology and cultural relativism,
descriptive, lexical, and factorial trait psychology, and classical psychometrics.
Some of these traditions strive to escape body-mind interaction traps by
sticking to one or a few levels of abstraction close to the top and by con-
centrating more on the end product than on the processes leading up to it.
They pay a high price for keeping biological factors out of the abstract analysis,
however. As they remain satisfied with pure description or correlation between
surface factors, 'causation' accordingly becomes a possibility rather than a
certainty in the empirical sense. This critique applies to behaviorism as well,
despite its explicit intent to turn into an objective natural science. Correlation
is the major analytic tool here, and a build-up of hypothetic associations the
explanatory devices. However, only by opening the "black box" of the
organism, and then having to acknowledge that the tremendous individual
differences observed in brain structure and function mean something to
phenotypic behavior, could behaviorism ever transform into an experimental
natural science, but then it would no longer be behaviorism and its chara-
cteristic surface analysis would turn into a top-down or bottom-up approach.
Top-down analyses typically take point of departure in some phenotypic
surface observation, and then strive to identify the most likely biological
candidates for explaining the observation. Many good examples of studies of
the biological underpinnings of intelligence and personality are provided in
this volume. Bottom-up analyses typically take point of departure in some kind
of manipulation of one or more biological parameters and then use phenotypic
behavior as a dependent measure. Of course, top-down and bottom-up
analyses can be combined or explored serially. They do not necessarily run into
serious body-mind problems if they stick to only a few neighboring levels of
explanation close to the top or bottom, that is, keep within compatible levels of
explanation. All too often, however, psychological analyses operate with a wide
spectrum of variables of very different degrees of abstraction, ranging from
genes and biochemistry over metabolism to brain structure and function, to
self, motives, desires, will-power, intentions, goals, theories and attitudes, to
individual phenotypic measurements, over social norms to collective behavior
and society in general. It all looks very convincing when presented in textbooks,
but what it really boils down to in the end, is a futile exercise in trying to
camouflage unavoidable Rylean category errors with truly catastrophic effects
for the causal analysis. Modern psychology, based as it is to a large extent on
such clever exercises of rhetoric, may actually harm more than it benefits the
scientific study of human nature by deflecting the focus away from the
essentials in the causal chain of events (Nyborg, 1994a).
Nature-nurture models may be seen as much more precise tools, but they
represent just another case of dualist rhetoric. They are typically built around
Fisher's analysis of variance model, and presume that independent proportions
568 Epilog

of genetic and environmental variances can be identified and added up linearly

to explain 100% phenotypic variance. While quite useful at the descriptive
level, such models are causally as empty as classical psychometrics, because
their focus is on individual differences around population averages. They
thereby statistically average across person-specific within-group differences in
actual causal agents and mechanisms (Nyborg, 1987, 1990, 1994a). Unidenti-
fied genes are presumed to interact through unknown mechanisms with highly
abstract and usually intuitively defined environmental components such as
rearing, norms, and social interaction without any consideration for the
biological locus of action except in terms of an unspecified mind or brain. The
extended use of reified superorganismic concepts makes nature-nurture
models basically dualistic. Molecular genetics might provide a much needed
differential cause-effect perspective on the nature-nurture interaction, but
then we are no longer talking about the traditional nature-nurture models in
behavior genetics, but an all-bottom type of analysis. The all-bottom approach
is described at the end of the epilog.

2.3 Hans Eysenck's position

Eysenck spotted many of the dangers long ago (e.g., Eysenck, 1952,1960,1967,
1970, 1979, 1983, 1985, 1996), and proposed a detailed program for suc-
cessfully proceeding beyond surface analyses based on psychometrics, correl-
ations, and factor analysis. These analyses have performed well in the past,
according to Eysenck, but they do not bring us below the descriptive level.
Eysenck therefore wants to combine the correlational with the experimental
tradition, in order for psychology to finally attain proper unitary scientific
stature. In this he follows the advice of Cronbach, addressing the American
Psychological Association in 1957. But Eysenck stresses again and again the
critical importance of establishing a good (cognitive?) theory before starting an
experiment. This intent is seen most clearly in the introduction to his book
Genius: The Natural History of Creativity (Eysenck, 1995, p. 1) where first of all
he quotes W. L. Bragg: "The important thing in science is not so much to
obtain new facts as to discover new ways of thinking about them." Too often
psychologists search for "facts" without stating a prior theory. But then again,
Eysenck agrees that: "One's knowledge of science begins when he can measure
what he is speaking about, and express it in numbers" (Lord Kelvin, cited in
Eysenck, 1995, p. 4).
Eysenck wants to re-construct psychology on the basis of several different
traditions. To Danzinger's (1990) three models of research: "The experimental
(Wundt), the psychometric (Gallon) and the clinical (Kraepelin)," Eysenck
would like to add a fourth: "The psychophysiological-genetic approach
(Helmholtz)."
 Psychology as science 569

With the correlations approach, the new Eysenckian psychology has an

explicit focus on the individual differences tradition coming from differential
psychology. It certainly is no coincidence that Eysenck was one of the founding
fathers of the International Society for the Study of Individual Differences, and
also the founder and co-editor-in-chief of the prestigious journal Personality
and Individual Differences since it went to print. With his differential
perspective, it is worth noting that Eysenck is not a particularly great admirer
of the idiographic approach:
If a person is unique we cannot study him scientifically, because we cannot measure
his unique aspects, or compare him with others. We cannot even prove that person-
ality is unique, because that would involve measurement, which is explicitly
condemned as disregarding uniqueness! Indeed, we would have to abandon all
psychological terms and concepts which allow us to compare individuals; ... This
whole approach ... leads to a completely non-scientific approach to the study of
creativity and personality. (Eysenck, 1995, p 7)

Eysenck here criticizes Allport's (e.g., 1965) preference for studying the
unique personality of the single individual. Allport, for his argument, objected
strongly when Eysenck (1952) stated that: "To the scientist, the unique
individual is simply the point of intersection of a number of quantitative
variables." Allport found that the fully legitimate interest in analyzing the
mutual interdependence of part-systems within the whole system of
personality, was threatened by the analysis of separate dimensions whereby
many persons may be compared.
A fourth step in a fully developed Eysenckian psychology is its transformation
into the ranks of the natural sciences (H. J. Eysenck and M. W. Eysenck, 1989);
here he sides with many other notabilities, including Watson and Skinner (and
William James, see later). The "fuzziness" of psychological concepts such as
intelligence, creativity, and personality is no hindrance for such a move, according
to Eysenck. Psychological concepts are no more fuzzy than physical concepts like
gravitation, electricity, or metal. Gravitation and temperature can, for example,
be defined in various ways as can intelligence, but that renders none of these
concepts useless, even if it awakens caution. Though the notion of "action at a
distance" appeared absurd even to Newton himself, he used it well and so do we
today (but now sided by at least two other very different definitions). Neither is
psychology, methodologically speaking, much worse off than physics.
Temperature measured by resistance and by expansion of metal give scales
that differ somewhat in various ranges. So do different measures of intelligence
and personality. The point is that concepts in any science should be judged in
terms of their usefulness or uselessness, says Eysenck. Then he stresses once
again, that we first have to elaborate a useful theory, or no functional laws are
likely to be found.
570 Epilog

In a recent special review of Hormones, Sex and Society: The Science of

Physicology (Nyborg, 1994a), Eysenck outlined in a very broad sense his view
on the theoretical underpinnings of psychology (Eysenck, 1996). He sees the
field of psychology as suspended within a triangle, cornered by Titchenerian
psychology, behaviorism, and reductionism/materialism. Titchener (1909,
1912), it will be remembered, was inspired by J. S. Mill's (1865) notion that
sensations are fugitive and temporary. However, there are "permanent possi-
bilities of sensations" that last. Titchener developed this idea into a two-stage
context theory to account for meaning. Seeing a face for the first time provides
little meaning. Seeing it again, and adding the context to the visual core, such
as previous contextual visual or auditory images, and the face provides
meaning. With repeated perception and habituation, the context may drop off
and the meaning takes on a purely physiological form. It is now embedded and
unconscious. Titchener was not alone with this view. Also George Berkeley
(1709) thought that to create meaning, at least two sensations (images) are
needed, but with later use, one sensation might suffice. Eysenck is actually
making much practical use of Titchener's context theory of meaning, which can
be seen quite clearly in his recent Genius book. Eysenck was also inspired by
Titchener's dimensional view of mind. It is thus not too difficult to see why
Eysenck finds that Titchenerian psychology ought to be one of the three
cornerstones in psychology.
Commenting on (what he sees as) my philosophical materialist-reductionist
stance (I am neither a reductionist nor a materialist, and I most certainly do
not afford the luxury of having a philosophy, as will be apparent later!),
Eysenck notes that it is entirely a matter of personal philosophy whether a
particular researcher feels more at home in Titchener's corner, chooses the
behaviorist angle, or experiences a reductionist itch towards materialism.
There simply is not yet enough hard evidence in store today, he adds, to make a
qualified choice among these alternatives.
The above description may account in broad outlines for Eysenck's view of
psychology at the brink of the twenty-first century. At least, it hopefully
provides the reader with an idea of how far Eysenck is prepared to go in
denouncing classical psychological views in future examinations of human
nature.

2.4 Unended quests in Eysenckian psychology

Eysenck has without question had a colossal impact on the development of
twentieth-century psychology and psychiatry. In the following I will first ultra-
briefly comment on selected highlights of his impact, and then focus in more
detail on unended quests.
 Psychology as science 571

Many chapters in this volume testify to Eysenck's genius in identifying likely

distal evolutionary and proximal biological underpinnings of consequential
phenotypic behavior, even in early times when little empirical evidence could
be mustered in defense of the view. This legendary foresight may, in part,
explain why Eysenck is the most cited psychologist, and next to Marx and
Freud (what strange bedfellows!) is the most cited person ever. If citation
means importance, then Eysenck is extremely important. He has set a heavy
mark on many directions in psychology. He has exposed the fallacies of
psychoanalysis, and opened many people's eyes to the importance of a proper
scientific approach to personality and intelligence. He has formed the course of
behavior therapy, and changed the way we look at psychopathology. He has
emphasized the use of Pavlovian principles, so that concepts of conditioning,
arousal, and cognitive inhibition found a safe place in Western psychology, and
he has done so with admirable clarity and at times also with surprising force.
When I once commented on the latter aspect, he just laughed and said "I don't
mind blowing fresh winds through the dusty halls of academia." The contri-
butors to the present volume have done their best to document Eysenck's
influence in their areas, by illustrating how it guided research, clarified or
provoked counterattack.
On balance it is only fair to mention that several of the chapters also air
concern about unended quests in psychology. Although Zuckerman readily
admits his profound debt to Eysenckian personality theory, which he finds
provides a much-needed bridge between genotype and behavior via
biochemistry and neurologic intermediaries, he nevertheless finds it appro-
priate to present an "alternative-5" model, and to suggest that his own
sensation-seeking scale taps deeply into Eysenck's psychoticism dimension.
Eysenck and he continue to discuss this matter. Chris Brand finds it
problematic to settle for just three Eysenckian personality dimensions.
Perhaps very intelligent people have a more differentiated personality than
the less intelligent. For this and other reasons, Brand therefore settles for an
alternative "Big 6" solution. Eysenck is not particularly happy with this, and so
the discussion about the dimensioning of personality goes on. Brand further
wonders whether Eysenck's search for psychological underpinnings will bring
anything useful that previous mechanistic approaches couldn't. He suggests
that dimensional personality variation may be better accounted for in terms of
individual dynamics, purpose, and biological function. For reasons given in
chapter 20 and below I fear that to explain anything in terms of "purpose" is
another dead end.
Gray and the very active group around him are not too happy with Eysenck's
interpretation of his own E and N axes, so they rotated them and provided a
different causal interpretation. Eysenck in turn, is not too happy with the re-
interpretation. A related problem here is the fact, that the intensive research
series by Gray and others demonstrated inconsistencies in their own approach
572 Epilog

and, still worse, the group sees no easy solution to these problems. Strelau and
Zawadski ask whether Eysenck is correct in assuming that his personality
dimensions equal temperament. They found that E and N correlate with
almost all temperamental scores, but also that P relates to sensation seeking
only, a finding that interests Zuckerman. They suggest that P differs
fundamentally from E and N, and perhaps reflects a temperamental
disposition to inhibit impulses. Nias carefully inspects the controversial
evidence on cancer-personality and nutrition-intelligence relationships. Over-
all, he finds that the observations made by Eysenck and others in these areas
are promising, but he also stresses the need for good replication studies.
In light of the ongoing discussion of the status of the P dimension in
Eysenck's personality theory it is interesting to note Sybil Eysenck's warning in
chapter 6, that leaving P (and L!) out of consideration will seriously damage
the true picture of personality. Reviewing the evidence for Eysenck's biosocial
theory of crime, Raine finds that large parts of the theory are intact and well
after all these years. He wonders about its future, however, and ask whether
the problem of growing crime and violence would require a modification of the
theory. Methodologically, Eysenck's biosocial theory could benefit from
molecular genetics and discordant twin studies. New imaging techniques
could further increase our knowledge of brain dysfunction in arousal,
conditioning, and emotional regulation, and permit us to go beyond skin
conductance and heart-rate measures. It concerns Gudjonsson that a high P
score does not always correctly identify persistent criminals, and that high P
scorers actually form only a small part of the criminal population. He further
finds that Eysenck's theory of crime has had more success with research into
the causal basis of crime than with attempts to prevent it. Wilson notes that,
whereas Eysenck's personality theory seems to account reasonably well for
essential aspects of sexual behavior, it encounters problems in explaining
sexual conditioning. Revelle concludes part I by reviewing the evidence on how
impulsivity relates to extraversion. He finds this task difficult, in part because
Eysenck changed his view over time, in part because Gray (and others) have
suggested alternative interpretations that Eysenck does not endorse. There are
also problems with how to best measure impulsivity. Revelle finds that
impulsivity is very important for the understanding of individual differences,
but it has not yet found its proper place in multidimensional personality theory.
As in part I, the authors of part II fully acknowledge Eysenck's tremendous
influence on the development of their area of specialization. As in part I they
report on remaining problems. Jensen underlines an important distinction in
research on intelligence by asking: Is reaction time a function of higher-level
cognitive processes?; or Can the speed and efficiency of neural processes
explain the observed covariance between reaction time and Spearman g? We
simply do not know, and Jensen finds that only further brain research will
provide the answer. Vernon finds that genes explain much of the phenotypic
 Psychology as science 573

covariance between intelligence and personality, but also that is time now to
move in new directions in behavioral genetics. Lynn notes, in the light of many
later studies, that Eysenck surely was right when he changed his early
environmentalist view and suggested that race differences in intelligence
probably reflect a genetic component. We are now concerned with the
important question of how the genetic component came into being in the first
place. Lynn proposes a geo-climatic theory. In dealing with the question of
intelligence and information processing, Deary notes that attempts to identify
the basic cognitive processes, primitive or raw brain processes, have so far been
largely in vain. Neither does he find much success for Eysenck's idea of a
unification of the differential/correlational with the experimental/cognitive
approach. Brody notes, in his attempt to determine the degree to which
environmental intervention affects intelligence, that Eysenck's authoritative
view that our biology predestines to a large extent our future seems basically
correct. That being empirically established by now, our next task is, according
to Brody, to examine in intimate detail the biological components behind
intelligence.
In part III, Irene Martin follows up on the vast importance of Eysenckian
personality theory, but she also notes that research on human eyelid con-
ditioning has not benefited much from the theory, and that there might be a
problem with whether high P/high E-low P/low E differences can be explained
in terms of impulsivity. Theoretical terms such as inhibition and excitation are
today used mainly as explanatory concepts, arousal has resisted attempts at
definition as has its links to attention, and there are not many bridges between
contemporary conditioning theories and therapy. Martin would like to see
further clarification of the physiological underpinnings of E/I and N, as she
finds Eysenck's notion of a factor of conditionability that transcends specific
response systems untenable.
The concepts of inhibition and excitation are now used as descriptive
concepts in cognitive research as a basic mechanism for fine-tuning cognitive
processes, but their precise role remains uncertain. She finally regrets that
conditioning studies often fail to take individual differences into account.
Claridge inquires into the rather open question of how to visualize the
biological elements common to neurotic anxiety-based high arousal states (N)
and the opposite "extraverted" (E) disorders. Like others, Claridge expresses
some concern about the metrics of the P scale and about how psychoticism has
been conceptualized—not so much from the personality theory side but from a
clinical point of view. He thus wonders whether the assumed close association
between aggression and schizophrenia in a dimensional scheme might be
looser than postulated. Perhaps part of a more modest relationship can be
explained by other factors? However, Claridge does not doubt for a moment
that P remains highly relevant for our understanding of serious mental illness,
though perhaps not with the status of a primary etiological factor in psychosis.
574 Epilog

It could actually be crucial in determining actual psychotic breakdown. In his

detailed review of the psychophysics and psychophysiology of extraversion and
arousal Stelmack finds some support for the view that introverts are more
sensitive to physical stimulation than are extraverts, and that differences in
extraversion involve individual differences in fundamental motor mechanisms,
quite as suggested by Eysenck. However, he finds little evidence for the notion
that individual differences in cortical arousal actually determine these effects.
Perhaps differences in extraversion involve peripheral brain stem and spinal
motoneuronal processes rather than central cortical arousal mechanisms,
Stelmack asks. But even if dopaminergic activity modulates sensory input and
response output, Eysenck's original conception is really not far from the mark.
Rushton, in his discussion of (im)pure genius and its relationship to
psychoticism and intelligence, expresses skepticism about the repeated
reassurances that creativity only correlates with IQ up to 120, and argues
that many of the most creative scientific disciplines today most likely require a
much higher IQ for success. Other evidence suggests that IQ correlates
significantly not only with the complexity of a task but also with achievement
within an occupation. Rushton finds that the ordinary term "motivation" may
be too self-willed to explain the almost obsessive-compulsive behavior of many
geniuses. Perhaps they "get high" more easily than others in a fashion that
looks very much like being under the influence of stimulant drugs. Perhaps the
urge of a genius is better described in terms of acting out a unique value
system, "super-ego" or a "concern for excellence" than in terms of the
personality traits at the high end of the psychoticism scale, Rushton suggests.
Nyborg notes, while discussing details of Eysenck's creativity theory that the
particular genes made partly responsible for shaping the components behind
creativity are not yet identified, and that dopamine more likely is an important
covariate than the key physiological variable in creativity and psychoses.
Moreover, a strong dopamine hypothesis runs into difficulties in explaining the
marked sex-related differences in creativity. Nyborg notes that the role of P in
creativity and psychoses might be more complex than anticipated. Finally, he
finds that multiplication of inferred psychological with measured biological
concentration factors leads to serious problems with the interpretation of the
end product. What does it mean in causal terms to multiply high dopamine
with ego-strength?
Even though Eysenck has not himself published much on occupational and
organizational psychology, his personality theory had affected many working in
these areas, according to Adrian Furnham, and led to the construction of the
Eysenck Personality Profiler measuring 21 primary factors in addition to the
super factors PEN. Furnham finds that, in general, the extant research from
classical personality theory on occupational/organizational variables has been
largely disappointing, and that the few interesting results are rarely followed up
 Psychology as science 575

and the theoretical implications exploited. It is to be hoped that Eysenck's

theories would in the future make the occupational psychology and
organizations behavior literature benefit from each other.
Suitbert Ertel explains Eysenck's interests in sunspot-related bursts of
creativity, and notes that anomalies in solar activity do covary with dis-
continuities in human cultural history. At the same time Ertel cautions that the
explanation for this is presently quite speculative. He hopes that our present
ignorance about helio-dependent effects on creativity, perhaps mediated via an
underlying neuropsychological mechanism, will soon diminish, but refrains
wisely from expecting surprising advances in the field. Dean, Nias, and French
review Eysenck's involvement with graphology, astrology, and parapsychology,
and notes that each area has a solid core of testable ideas. Actual tests suggest,
however, that effect sizes for graphology are too small to be useful, astrology
shows effect sizes that probably can be explained by nonastrological effects,
and nobody has so far claimed the million dollars in prizes for the conclusive
proof of psi.
Summarizing this brief exposure of unended quests, there seems to be fairly
widespread disagreement about the number of personality dimensions as well
as their causal interpretation; some find that the suggested psychological
underpinnings of both personality and intelligence meet problems with
interpretation and empirical foundation; almost all agree that the biological
underpinnings of both need further elaboration and clarification. Another way
to phrase this is that there is a growing consensus—including Eysenck
himself—that essential areas in contemporary psychology face problems with
explanation in terms of psychological concept, and that future research needs
to focus more precisely on the physiological basis of human nature. Translated
into the Eysenckian triangular view of psychology, this means that many
leading scientists in the areas of intelligence and personality are now on the
move towards the reductionist/materialist corner in the triangular space,
demarcated by Titchenerian psychology and behaviorism in the other two
corners. At the same time, few seem ready to entirely skip psychological or
cognitive explanations in one form or another.
This ambivalence creates an interesting situation. On the one side, there is
no doubt that further work on the psychological side will lead to increased
theoretical sophistication and, perhaps, also to methodological elaboration of
top-down and bottom-up approaches. At the same time, there is a real danger
that further elaboration of psychological theory will intensify the Babylonian
body-mind confusion, if only at an ever higher level.
Given that this characterization of an increasingly difficult situation is not
totally overwrought, it is perhaps time to stop, and try to resist further
temptations to explain human nature in terms of mentalism and anthro-
pocentrism. The next section briefly discusses the outline for a program
designed specifically to study (human) nature without mentalist connotations,
576 Epilog

body-mind confusion, and anthropocentrism. Space limitations leave room for

only a simplified account, but even this probably suffices to sicken some of the
happy psychologists and philosophers who faithfully subscribe to mentalism
and love to chase reductionists.

3. THE TWENTY-FIRST-CENTURY SCIENCE OF HUMAN NATURE

3.1 Introduction
A new "mindless" program would see the body-mind schism as the principal
stumbling stone for developing a proper science of human nature. It
accordingly dispenses entirely with all high-level Platonic abstractions and
the dualism they encompass. However, multilevel analysis is not per se seen as
the problem. As long as the levels do not refer to fundamentally incompatible
categories, intractable category errors can be avoided (Ryle, 1980). With
abstraction reduced to an absolute minimum, the new research program can be
found near the bottom of Figure 25.1.
Eysenck has, as usual, long since spotted the new writing on the wall, as have
most of the contributors to this book. It is equally obvious that neither Eysenck
nor most of the fellow contributors are ready yet to jump on the exclusively
materialist/reductionist bandwagon. Perhaps this is a wise decision. I,
nevertheless, found the following three pieces of information sufficiently
important to prepare me for the jump. I landed not in the philosophically
defined materialist wagon, but in a type of molecular analysis of (human)
nature that philosophers probably would call an expansionist position, but what
philosophers say is in general rather immaterial!
First, more than two-thousand years of almost desperate search by the best brains for
mental stuff has brought us nothing but postulates. This is, of course, no proof that it
does not exist! However, an extensively researched phenomena stops serving as a
useful heuristics when century after century turns without the slightest sign of proper
operationalization or verification, and it becomes an increasingly less likely candidate
for twenty-first-century empirical research. This alone was sufficient reason for me to
jump off the intuitively defined mentalistic-cognitive-philosophical bandwagon.
Second, Watson and Crick's success with their mid-century attempt to break the
genetic code for life was a powerful demonstration that even highly complex pheno-
mena such as "life" are fully amenable to exclusively natural-science analysis.
Third, the recent breathtaking advances in biochemical and brain sciences make it
obvious that new technologies already now provide fairly precise answers to questions
which were previously addressed with intuitively based speculations and the use of
philosophical and psychological tools. The new techniques address the brain directly
and the results do not depend on philosophical abstraction.
In other words, any scientific twenty-first-century program for the study of
human nature should start by drawing the inevitable conclusion that mentalism
has failed, and then give it back to the armchair philosophers without
 Psychology as science 577

gratitude! Second, any new program should try and combine the lesson from
the successful demonstration that all different life forms reflect different
combinations of just four bases, with the lessons emanating from the recent
explosive methodological progress in physics, chemistry, and brain imaging
techniques. Together these lessons will provide a fertile soil for a new
generation of cross-disciplinary scientists by allowing them to harvest genuine
empirical fruits in formerly so-enchanted gardens. No longer burdened by the
futile task of circumventing or hiding anachronistic body-mind traps, these
scientists would be free to examine how matter interacts with matter in humans
as everywhere else in the universe, in accordance with a few simple natural-
science principles. Of course, such a move has to be closely monitored experi-
mentally by the uncompromising application of step-by-step testable models
operating at a coherent level of analysis. The big bonus is, that all the lofty talk
about complementary effects between a qualitatively different subjective
observer and the objective world would boil down to empirically addressable
questions about what interacts with what at the material level, and only that.
Obviously, in this approach the term "model" would refer to particular
physico-chemical brain states.
One research program designed to accomplish this is called physicology.
This molecular all-bottom approach was named physicology, because all
explanations based on abstractions like psyche, ego, or cognition and high-level
superorganismic concepts are substituted with references to their physico-
chemical underpinnings. The physicological research program is presented in
detail in (Nyborg, 1994a, b, 1997a, b). Here only a few aspects will be high-
lighted, and then only as they relate to relevant aspects of Eysenckian
psychology.

3.2 The use of theory in psychology and physicology

Eysenck insists, as mentioned before, that without a good (cognitive?) theory,
it will be impossible to find functional laws. A physicologist would question
this, by demonstrating that such a task basically needs only two a priori
assumptions (read: molecular states): (1) That molecules have differential
stereotaxic affinity; and (2) that the affinity-related flow of energy in space-
time coordinates defines the developmental characteristics and behavior of any
system, irrespective of molecular complexity and whether it is carbon-based
(i.e., organic) or not (i.e., inorganic).
A simple case illustrates the point. Let three molecules contact each other,
randomly or forced. Given that two of the molecules have high stereotaxic
affinity for each other, they might form a new species of chemical behaving in
ways that differ from the previous behavior of the single molecules, while all
the time leaving the third molecule with a different affinity untouched. The
differentiation in this very simple system, including the birth of new
578 Epilog

functionality can (1) be described entirely in physical and chemical terms, (2)
be analyzed by ordinary natural-science methods, (3) be extended ad libitum to
systemic behavior of any complexity, including humans or stars, and (4) it does
not require any a priori theory to observe how meeting molecules with
different affinities bring different types of order into an otherwise homogenous
or chaotic world. In fact, physicology dispenses entirely with abstract concepts
such as theory, understanding, meaning, and purpose. Such anthropocentric
terms do not apply to molecules, whereas affinity, space-time position,
concentration, action, reaction, conformation, transcription, and metabolism
do. We only have an untrustworthy nonempirical philosopher's pessimistic
prediction, that this all-bottom analysis will never suffice to describe the nuts
and bolts of human nature and society.

3.3 The three analytic windows in physicology

To start assessing whether the two a priori mentioned above are not only
necessary but also sufficient for the study of human nature, the physicological
analysis begins with the opening of one or more of three windows: The
intrasystemic, the intersystemic, and the extrasystemic (for details, see text and
Figure 20.3 in chapter 20, or Nyborg, 1994a). The material basis for various
mentalist and superorganismic reference concepts can then be examined
through these windows by examining their most likely systemic physico-
chemical addresses, the interaction of these with activities on other physico-
chemical addresses in the vicinity, and the consequences of all this for, say,
phenotypic behavior. Behavior is, in this framework, seen as a special case of
the global molecular transport of complex (carbon-based) systems (like
humans), or parts thereof, in space-time coordinates.
The physicological analysis through the intrasystemic window presumes that
consciousness refers to physico-chemical rather than cognitive processes;
intelligence, personality, and symbolism to intrasystemic metabolism rather
than a flow of ideas; meaning to molecular changes instead of the use of logic
or attribution of emotional importance; cognition to synaptic activity;
premonitions to the state of physical parameters, and philosophy corresponds
to molecular flow-patterns.
The intersystemic window is reserved for a special case of extrasystemic
analysis, namely, the molecular correlates of so-called social interaction. Social
interaction is defined as systematic exchanges, not of signs, attitudes, or
meanings, but of physical stimulus patterns. Pedagogy is the more or less
systematic environmental realization of molecular options within systemic
physico-chemical constraints, not limitless accumulation of abstract
instructions, norms, or culture conveyed by significant others. Love refers to
a special set of coordinated adjustments in gonadal hormone parameters in
two or more individuals, ultimately in the evolutionary service of another
 Psychology as science 579

entirely physico-chemical phenomenon: Reproduction. This is not exactly the

way romantics see it, but they may have to admit that this arrangement fits
perfectly in a natural-science view of our evolutionary past as a series of
entirely physico-chemically coordinated arrangements. At least the physico-
logical analyses need not be pestered by unclear anthropocentric ideas of
mysterious mental qualities emerging during evolution to raise humans to a
status close to God.
The extrasystemic window focuses on all nonsocial external physico-chemical
stimulants, some of which may have an impact on intrasystemic physico-
chemical parameters. An example is the hormonal exchanges between the
mother and her child in the womb. This prenatal arrangement can dramatically
influence the body or brain development of children. Moreover, a multitude of
variations in the mother's entirely physico-chemical environment may alter her
body and brain chemistry, which in turn may affect her unborn child. This
illustrates that the physicological analysis does not operate on the previous
widespread assumption of a decisive inner-outer distinction with the skin as
the border between subject and object, but is rather concerned with dynamic
molecular interactions, the foci of which may be examined through different
analytic windows.
The analysis of cultural differences also belongs to the extrasystemic
window, and this is not as surprising at it may first seem. Culture-related
similarities in behavior (themselves molecular phenomena) reflect, according
to physicology, a certain degree of molecular commonality in geographically
defined groups of people. These commonalities came about as a function of
selection among many different molecular constellations. Only those constel-
lations that were compatible with economic adaptation to their particular
ecological niche during evolution survived, had a reproductive advantage, and
were able to raise competent children capable of surviving in the harsh
competition in their generation. There is no place in a causal physicological
analysis for unclear notions of reified local prescriptions, cultural stereotypes,
or other superorganismic concepts with postulated effects on behavior.

3.4 Uniqueness, causality, and the science of human nature

Eysenck holds, as mentioned, the opinion that a unique person cannot be
studied scientifically, because there would be no standard with which to
compare this uniqueness. We would have to abandon all psychological terms
and concepts which allow us to compare individuals, and no functional laws
could be derived.
In contrast, uniqueness presents no particular problem in the physicological
analysis. The meaningfulness and explanatory value of psychological terms is
questioned anyway, and all functional laws can, in physicology, be derived from
one simple basic molecular characteristic, the functionality of which does not
580 Epilog

depend on theory. Psychological terms, concepts, theory, and philosophy serve

in this system only as convenient descriptive shorthand references in need of
physico-chemical addresses in space-time-phase coordinates.
The important a priori for the physicological analysis is, as said before,
stereotaxic affinity that primarily harnesses what molecules can and will do
when they meet. It is affinity that reduces entropy by self-organization of
molecules into body and brain structures. People differ substantially in which
molecules meet where, when, and under what circumstances, during develop-
ment and adulthood. It is this molecular variability that determines individual
differences in internal structures and functioning and, accordingly, in
phenotypic behavior at large. We seem to be talking about molecular con-
tinua here. It has to be remembered, however, that nonlinearity is the rule
rather than the exception in molecular cause-effect relationships (e.g., Nyborg,
1994a, 1997a, and see later), and this dissolves the problem in traditional
(linear?) psychological theory with understanding how to analyze individuals
getting "unique" scores. A "unique" phenotype is a natural causal con-
sequence of a unique molecular constitution but it, nevertheless, conforms fully
to the same law that defines "modal" development, including individual
differences in body, brain, intelligence, personality, and death.
The physicological view is, to press the point slightly, that there is no
scientific detour around uniqueness; this is the only direct way to reveal the
true causal basis of human nature. In this, physicology actually agrees with
Allport's (e.g., 1965) strong advocacy for the study of uniqueness, and
disagrees with Eysenck's strong condemnation. However, with respect to
causal analysis, physicology is more in line with Eysenck's emphasis on
physiological and neural factors than with Allport's somewhat imprecise stress
on "dynamic" psychological factors.
Methodologically, the task of generalization in physicology goes through a
series of comparisons of individual molecular constellations to see to what
extent they represent common characteristics (Nyborg, 1977), so we have to
start with individuals (unique or not). This is so, because the major weakness of
the usual group averaging approach and the study of individual variations
around the mean is that they do not allow for closer examination of the causal
agents and mechanisms underpinning phenotypic characteristics. The
averaging process does not allow for control of whether different causal
agents or mechanisms result in similar phenotypic scores, neither does it reveal
whether different phenotypic expressions were caused by similar agents or
mechanism. As everything is averaged, the individual with all its internal
characteristics disappears in the process. What we are talking about here is the
question of using the individual differences versus the different individuals
approach. Unfortunately, the detrimental consequences for the exact causal
 Psychology as science 581

analysis of choosing the individual differences approach are often overlooked

in nature-nurture studies (Nyborg, 1987, 1990a,b) as well as in psychometrics
(Nyborg & Sommerlund, 1992).
To summarize, Eysenck finds that the unique person cannot be studied
scientifically in psychology, whereas physicology finds that there can be no
proper causal study without focusing on the individual, unique or not. We must
insure that we have correctly identified the relevant causal agents and
mechanisms in each individual before we begin to generalize across individuals
to human nature. I called this the "idiothetic" approach (Nyborg, 1994a,
p. 164), because it combines the idiographic with the nomothetic approach.
The idiothetic approach may in fact be the only scientifically acceptable way to
causally connect DNA, over body, brain, intellectual and personality develop-
ment, to society, without having to average across different causal events,
mechanisms, and effects and thereby obtaining an anonymous average that
may neither fit any particular individual in the group nor say anything in
particular about human nature.

3.5 Methodology
On the methodological front, Eysenck recommends a unification of the
correlational/descriptive with the experimental/cognitive method. As the
physicological view on what determines human nature differs fundamentally
from the psychological view, it is only natural that physicology also calls for
other methods. A few examples may illustrate this point.
Nerve cell membranes can be defined either at the cellular level, or as
conglomerates of associated molecules "frozen" in space-time coordinates in
accordance with their specific affinities (e.g., Nyborg, 1997b). Organs can be
defined in terms of their overall function, or as mass-assemblies of molecules.
Neurotransmitters can be defined as chemical species, or seen as molecules
associated in a fluid state. As will be known by now, in physicology it is dif-
ferences in molecular cohesion that is the common principle behind all these
seemingly very different bodily and brain structural and functional mani-
festations. The molecular focus allows the physicological analysis to dissolve the
usual sharp distinction between structure and function at the bottom level.
Another consequence is that body and brain structures and functions are defined
not so much in terms of permanent traits as by more or less permanent molecular
states with all sorts of gradual temporal transitions and changes as the neigh-
boring molecular circumstances dictate. Plasticity becomes an option rather than
an enemy in structural-organizational terms. The dynamics pave the way in
physicology for defining the brain as a fairly unstable molecular system constantly
at the brink of sudden change. Instability or changes in, say, the permeability of a
nerve cell membrane can be mediated by sudden endogenous coordination of
otherwise chaotic firings in channel proteins building up energy, or by stimulus-
582 Epilog

phase-locked or stimulus-related changes caused by exogenous import of energy.

In fact, the major technical difference between an educable brain and a stone is
the difference in molecular stability. As the method must suit the analytic task at
hand, a physicologist would obviously prefer natural-science methods to handle
such molecular mass-interactions. These methods make us able to check the
causal chain of molecular events behind observations in psychology, anthro-
pology, or sociology.
Biological actions and reactions are often, as said, of a nonlinear nature.
This complicates the analysis tremendously, but a fully developed physico-
logical analysis must be able to handle this (Nyborg, 1997b). Massive number-
crushing computers are needed for the task, as are numerical and graphical
simulations. Descriptive sequential analysis of massive molecular cascades of
interlocking events must be dealt with in appropriate ways. Pattern-recognition
algorithms have to be adapted and perfected. Existing parallel computation
algorithms of fluid dynamics will need adaptation to fit the task at hand, and
we need to attend to problems with shared and distributed memory parallel
computation of dynamic load balancing and parallel fluid-flow visualization,
preferable in the form of small "movies" to illustrate covariant changes over
time. Of course, none of the simulation techniques can substitute actual wet
experiments or real-time brain monitoring, but they may act as a valuable
supplement as we progress. No doubt, these complex computations and
simulations will greedily ask for hitherto unseen raw computer power, that will
make contemporary simulated nuclear explosions look simple.
The encouraging perspective is that it should, at least in principle, and
perhaps one fine day also in practice, be possible to simulate the molecular
machinery of a human being in dynamic interaction with its environment in
real time. No longer bothered with mysterious intervening variables and effects
of hypothetical constructs at a high level of abstraction, it should be possible to
go beyond "simple" mappings of the genome, and reconstruct the molecular
cascades of events that unfold in the space-time coordinates between DNA
and the physico-chemical environment. This would amount to nothing less
than finally coming to terms with human nature by demonstrating that an
individual is a material-interacting organizational part of nature rather than a
philosophical construction isolated from and elevated above a material world.
In that case the ancient body-mind problem would finally have withered
away after 2400 years of obsessive search for still more sophisticated abstract
constructs and references to a mind that might not be there. What remains
after the body-mind dust has settled are thus a number of practical problems.
They are complex, indeed, but nevertheless much easier to tackle than the
quicksand of dualism, from which there is no escape. The bad news is,
however, that there is not the slightest doubt that methodological imperfection
will for quite a while constitute a major bottleneck and be the worst enemy of
progress along these lines.
 Psychology as science 583

3.6 Reductionism, materialism, or expansionism?

Reductionism is clearly held in favor by Eysenck but he is, nevertheless, not
ready for the wholesale reduction of psychology. He supports a Helmholtzian
psychophysiological-genetic approach but not full-scale materialism. When he
discussed my position as materialistic/reductionistic, he added that I might
demur with the designation.
This caution is well placed, because physicology is not a philosophical
position in any practical sense of that term. It is a simple research program for
the study of what happens when molecules meet and some of them stick
together. Odd things happen when this occurs, indeed. People and stars are
examples of such events, but their coming into existence might seem odd only
because of our ignorance! Their ontogenesis has nothing to do with philosophy
and everything to do with stereotaxic affinity and changes in energy distri-
butions. This hardly counts as a philosophical position in the usual sense.
Critics sometimes say: As a physicologist you say that you have no philosophy?
That is then your philosophy! Such philosophizing friends do not seem to
realize that such word games mean nothing to a causal analysis, where the
proof is in the effect. Neither is materialism a suitable label for physicology.
There are many different materialistic positions, and none of them looks even
slightly like the ultra-minimalist position of physicology.
Eysenck sees physicology as representing a reductionist position, and this
requires a comment. The reductionism critique boils down to an ingenious
linguistic gadget, invented by philosophers to protect their precious abstract
conception of human nature. The logic of the reductionist critique allows philos-
ophers to accuse physicology of being a grotesque misrepresentation of the truly
elevated status of human nature. What they forget to say is that the reductionist
critique presumes an a priori existence of something above the material
dimensions to be reduced. This would be true if Plato's shadow on the cave wall
represented more than what was in the physical optics of the situation, more than
what met the eye, and more than what altered the molecular state of the
observer. Only in such cases would the reductionist criticism apply. However,
nobody has ever documented the existence of these extra and abstract represent-
ations (environmental or mental). The proof for the abstract mind is in the
postulate, intuitively understood and never documented. The reductionist criti-
que is therefore shooting itself in the foot, and cannot be taken seriously except in
circles where arguments precede evidence. This becomes all the more obvious in
an era where the brain sciences come closer and closer to answering pertinent
questions previously far beyond empirical reach.

3.7 Natural science and the study of human nature

In the physicological analysis, the role of theory and concept are much less
important than are answers to empirical questions about affinity and molecular
mass interaction. It, for example, does not matter greatly whether we associate
584 Epilog

one concept or another with the molecular processes of defeminization and

masculinization of a female fetus by androgens, or with the inverse cycling of
female verbal and visuo-spatial abilities as a function of monthly changes in
plasma estrogen. What's in a name? The important task would rather be to
monitor where sex hormones go in the organism; which receptors they induce
where in the organism; what happens when the hormone-receptor complex
enters the cell and is activated, and which genes in the nucleus change expres-
sion when high-affinity parts of the genes "steal" the hormone from the lower-
affinity receptor and begin or stop producing proteins as a result; where the
newly transcribed protein products go in body and brain tissues, what their
biological effect is there, and what all this means for the process we call sexual
differentiation. This is not just another clever name game of inventing hypo-
thetical constructs and intervening variables to explain what happens, but a
question of carefully monitoring in a step-by-step fashion the molecular cascades
of events in empirically verifiable ways with the use of natural science tools.
As mentioned in section 2.3, Eysenck, and many before him, have suggested
that psychology should ideally transform into a natural science, but nobody
found an easy way. Kohler (1960) remarked that most attempts to deal with the
body-mind problem tacitly accepted the existence of emergence—referring to
the assumption that when systems become very complicated, entirely new
forms of action would arise that were not valid on lower levels. Such a
discontinuity would preclude the transformation of psychology into a natural
science like physics, as would the assumption that values guide behavior.
Kohler indeed found it likely that an organism simply consists of special
configurations of cells, by which events are given particular directions, distri-
butions, localizations and so forth (ibid., p. 17). The system is open to absorb
energy from the outside, and all actions in the brain must, as a particular kind
of process, be known to natural science. Pepper (1960, p. 39) assumed that the
physiological body is a function of cells occupying a limited volume of the
space-time field, which would place an analysis of it solidly within the reach of
chemistry and physics. Putnam (1960, p. 175) found that the body-mind
problem is nothing but a different realization of the same set of logical and
linguistic issues, so it must be just as empty and just as verbal. Jacob (1982)
found that the ever-recurrent problem in the natural sciences is to get rid of
anthropomorphism, such as the purposive activity of man. He found that
"endowing the elementary particles that constitute matter with some kind of a
psyche does not help much, and the conclusion is inescapable that mind is a
product of brain organization in the same way that life is a product of
molecular organization" (p. 59). Jacob nevertheless found that the study of
man can neither be reduced to biology nor do without it (p. 62).
These eminent scientists saw, in other words, the many problems with
explanations above the brain, and recommended physical explanations instead.
Common to them was also, that they were unable to formulate the formidable
 Psychology as science 585

problems at hand in terms of stringent natural science methodology. William

James, while fiercely advocating psychology as a natural science, was
undoubtedly also the most pessimistic of them all. In 1890 he delivered the
manuscript for his two-volume Principles of Psychology to a publisher with the
following characteristics. It is "a loathsome, distended, tumefied, bloated,
dropsical mass, testifying ... that there is no such thing as a science of psychology
...." He indeed wanted psychology to be a natural science, but found that "the
waters of metaphysics leak at every joint...." He found it strange to " ... hear
people talk triumphantly of 'the New Psychology,' and write 'Histories of
Psychology,' when into the real elements and forces which the word covers not
the first glimpse of clear insight exists. A string of raw facts; a little gossip and
wrangle about opinions; a little classification and generalization on the mere
descriptive level; a strong prejudice that we have states of mind, and that our
brain conditions them: but not a single law in the sense in which physics shows
us laws, not a single proposition from which any consequence can causally be
deduced. We don't even know the terms between which the elementary laws
would obtain if we had them ... This is not science, it is only the hope for a
science."
To this I would like to add that now, more than a hundred years later, we
cannot even justifiably keep up the hope of turning psychology into a proper
natural science. Psychology is a hopeless science, and it will remain so until the
time when it has been totally liberated from all forms of abstract psyche,
cognitions and unconsciousness, symbolisms, mentalism, and anthropo-
centrism. But then it will no longer be psychology as we know it today, and
its name would be an oxymoron.
It is actually quite likely that molecular affinity represents James's dream of
a "single law in the sense of which physics show us laws" and from which
"consequences can causally be deduced." In that case, affinity will finally allow
the study of human nature to turn into an experimental science that can
address the complex phenomena observed in psychology, anthropology,
sociology, and philosophy in terms of stringent natural science methodology.
Hopefully, some economical brains will soon develop the tools desperately
needed for the proper handling of nonlinear dynamics of complex carbon-
based systems such as us. The implementation of the physicological research
program depends on this.

3.8 Concluding remarks

The transition of psychology to a molecular science like physicology will
undoubtedly be slow. It may take at least a generation or so. It is thus not
realistic to expect that the many happy entrepreneurs working on the psycho-
logical and philosophical tower of Babylon will readily admit that their whole
project was a terrible mistake from the beginning. As their talent and creativity
586 Epilog

lies basically in the verbal-hypothetical-philosophical domains, they cannot be

expected to just turn around and take up the study of molecular dynamics. The
social-humanistic and natural sciences seem to attract different levels of
intelligence as well as different personalities (see chapter 20). Planck probably
had a point when he said that scientific truth does not triumph because
opponents become convinced but rather because inventors of the old truths
die. So, a qualified guess is that for another long period of time we will see a
parallel race among, on the one side, psychologists and philosophers ascending
into new and higher realms of their home-made conceptusphere with armchair
theorizing as the elevator and, on the other side, an army of brain-oriented
natural scientists, bored with the many trivialities of modern high-tech physics
and eager to simulate, and perhaps in a distant future reconstruct, human
nature in all its fantastic physico-chemical variety in accordance with a few
evolutionarily speaking very conservative natural-science principles for what
molecules can do in terms of life, love, society, culture, and all other entirely
material matters.
Eysenck's plastic brain organization and dynamic synaptic functionality has,
in combination with stable and optimized personality parameters, allowed his
overall molecular organization to function extremely efficiently and to express
a prolonged tendency to go biological and experimental even in times where
this was reacted to by most other brain organizations as being even more
heretical than it is today. This activity has affected many contemporary brains
in such a way that they now waste less time with repetitively running narrow
(perhaps a predominantly left-hemisphere Wernicke and Brodmann area)
linguistic dead-end loops, and accordingly can spare energy for the more
demanding molecular activity that goes under the name of scientific activity.
According to physicology, scientific activity refers to the ways adaptive synaptic
activity in material brains comes systematically to terms with an entirely
material environmental reality, including people and society.

4. A PERSONAL REFLECTION

This book concludes with a personal reflection on Hans as friend (if I dare say
so to a man, who invited me to sleep in his daughter's bed; unfortunately,
Conny was away at the time, but Sybil at least took care of the electric blanket)
and as the institution that he also is. Hans is, in my opinion, to psychology what
Bach was to music. Both are towering figures that combine the best of their
time in elaborate compositions, marked by a developed sense for accuracy,
clarity, tempo, harmony, beauty, elegance, and a variety that often takes an
almost mathematical form. Both demonstrate stringent criteria for solid work,
both conclude an important period in history, and both point towards the
future. What more can one ask for?
 Psychology as science 587

Bach was acknowledged as a master by some in his period. However, as

Hans writes in his recent Genius book: High capacity is certainly no guarantee
for glory in general. The history of extreme creativity is full of appalling
examples where exactly the opposite was the case (see chapters 19 and 20, or
better, read the Genius book). When Bach one day quite unexpectedly
appeared at the gate of the castle of Friederich the Great of Prussia, the music-
loving monarch immediately terminated his exercise on the flute, his favorite
instrument, and asked the old master to improvise a piece of music, an art form
he practiced to perfection. The monarch very appropriately suggested
variations over the theme B-A-C-H. Johann Sebastian sat down and
composed wonderful variations over the theme—off his hands, just like that.
The monarch was grateful, thanked Bach, and told him that he could eat in the
kitchen. His Majesty then went off with the usual number of largely
insignificant court puppets to enjoy the waiting gala dinner in the hall.
Again, I find a remarkable similarity between Eysenck and Bach. Most
acknowledge their genius, but the glory often goes to less significant people.
Both are able to improvise over complex themes straight out of their brains
(see chapter 24). Both have had to accept that the halls of glory were often
occupied by lesser people. Only in recent years has a certain measure of glory
come Eysenck's way (again, see chapter 24). It is perhaps no coincidence that
Eysenck lets the first chapter in the Genius book (1995) begin with a citation of
Jonathan Swift:
When a true genius appears in the world,
you may know him by this sign, that the
dunces are all in confederacy against him.
It may warm your heart, Hans, to know that this book should be seen as a
token, that not everybody is in confederacy against you. In fact, there are many
who are grateful for your numerous brilliant improvisations over the years, and
who would like to see you preside in the seat of honor at the high table.

REFERENCES

Allport, G. W. (1965). Pattern and growth in personality. London: Clowes.

Berkeley, G. (1709). Essay toward a new theory of vision. In A. A. Luce & T. E. Jessop
(Eds.), The works of George Berkeley Bishop of Cloyne (Vol. 1) (pp. 143-239).
Toronto: Nelson, 1948.
Cronbach, L. J. (1957). The two disciplines of scientific psychology. American
Psychologist, 12, 671-684.
Danzinger, D. (1990). Constructing the subject: Historical origin of psychological
Research. Cambridge: Cambridge University Press.
Eysenck, H. J. (1952). The scientific study of personality. London: Routledge & Kegan
Paul.
588 Epilog

Eysenck, H. J. (1960). The place of theory in psychology. In: H. J. Eysenck (Ed.),

Experiments in personality, 2, 303-315.
Eysenck, H. J. (1967). The biological basis of personality. Springfield, IL: C. C. Thomas.
Eysenck, H. J. (1970). The structure of human personality. London: Methuen.
Eysenck, H. J. (1979). The structure and measurement of intelligence. New York:
Springer.
Eysenck, H. J. (1983). Is there a paradigm in personality research? Journal of Research
in Personality, 17, 369-376.
Eysenck, H. J. (1985). The place of theory in a world of facts. Annals of Theoretical
Psychology, 3, 17-114.
Eysenck, H. J. (1995). Genius: The natural history of creativity. Cambridge: Cambridge
University Press.
Eysenck, H. J. (1996). Special review of Helmuth Nyborg: Hormones, sex, and society:
The science of physicology. Personality and Individual Differences, 21(4), 631-632.
Eysenck, H. J. & Eysenck, M. W. (1989). Personality and individual differences: A natural
science approach. New York: Plenum.
Jacob, F. (1982). The possible and the actual. New York: Pantheon
James, W. (1969). Psychology (briefer course) (4th edn). Toronto: Macmillan.
Kohler, W. (1960). The mind-body problem. In: S. Hook (Ed.), Dimensions of mind.
New York: New York University Press.
Mill, J. S. (1865). An examination of Sir William Hamilton's philosophy. Cited in E. G.
Boring (1942). Sensation and perception in the history of experimental psychology. New
York: Appleton-Century-Crofts.
Nyborg, H. (1977). The rod-and-frame test and the field dependence dimension: Some
methodological, conceptual, and developmental considerations. Copenhagen: Dansk
Psykologisk Forlag.
Nyborg, H. (1987). Individual differences or different individuals? That is the question.
Behavioral and Brain Sciences, 10, 34-35.
Nyborg, H. (1990a). Good, bad, and ugly questions about heredity. Behavioral and Brain
Sciences, 13(1), 142-143.
Nyborg, H. (1990b) The nature of nature-nurture interactions. Behavioral Genetics
(Abstracts), 20(6), 738-739.
Nyborg, H. (1994a). Hormones, sex and society: The science of physicology. Westport, CT:
Praeger.
Nyborg, H. (1994b). The neuropsychology of sex-related differences in brain and
specific abilities: Hormones, developmental dynamics, and new paradigm. In P. E.
Vernon (Ed.), The handbook of the neuropsychology of individual differences (Chapter
4) (pp. 59-113). New York: Academic.
Nyborg, H. (1997a). Personality, psychology, and the molecular wave: Covariation of
genes with hormones, experience and traits. In: J. Bermudez, B. de Raad, A. M.
Preez, A. Sanchez-Elvira, & G. L. van Heck (Eds.), Vlth volume of personality
psychology in Europe. Tilburg, the Netherlands: Tilburg University Press (in press).
Nyborg, H. (1997b). Nonlinear gonadal hormone modulation of the brain and behavior
along the life cycle. In: U. Halbreich (Ed.), Gonadal hormones, sex, and behavior
(Chapter 9). New York: American Psychiatric Press (in press).
 Psychology as science 589

Nyborg, H. & Sommerlund, B. (1992). Spearman's g, the verbal-performance balance,

and brain processes: The Lynn-Vernon debate. Personality and Individual Differences,
13(11), 1253-1255.
Pepper, S. C. (1960). A neural-identity theory of mind. In S. Hook (Ed.), Dimensions of
mind. New York: New York University Press.
Putnam, H. (1960). A neural-identity theory of mind. In S. Hook (Ed.), Dimensions of
mind. New York: New York University Press.
Ryle, G. (1980). The concept of mind. New York: Penguin.
Titchener, E. B. (1909). Lectures on the experimental psychology of the thought-processes.
New York: Macmillan.
Titchener, E. B. (1912). Description vs. statement of meaning. American Journal of
Psychology, 23, 165-182.
Wundt, W. (1874, et seq.). Grundzuge der physiologischen Psychologie. Leipzig:
Engelmann.
This page intentionally left blank
Author index

Abel, L. 138 Apicella, A. 247

Abelson, A. R. 286 Apt, C. 174
Abraham, K. R. 375 Armitage, P. 517
Abraham, L. 480 Armstrong, C. 479
Achenback, T. M. 129 Arnett, P. 44, 45
Ackermann, H. 349, 355 Arvey, R. 480
Ackney, T. 278 Asberg, M. 190, 195, 196
Adlar, K. 471 Asimov, I. 497, 501
af Klinteberg, B. 198 Atkinson, R. C. 297
Ahadi, S. A. 195, 196, 198 Awad, A. G. 377
Al-Adawi, S. 47, 51, 52
Albert, R. S. 422 Bachelder, B. L. 287
Albutt, J. 380 Bachorowski, J. A. 195, 201
Alcock, J. E. 527, 531, 535 Badcock, C. 27
Alcorn, M. B. 294 Bagby, R. M. 196, 197
Alderton, D. L. 296 Baghurst, P. A. 315
Alexandra, R. 469 Baker, D. P. 131
Allard, I. 151, 154 Baker, L. A. 243
Allen, R. M. 166, 171 Baker, S. 395
Allport, F. H. 448, 569, 580 Balabun, M. T. 62
Allport, G. W. 68, 72, 82, 515 Ballenger, J. C. 397
Allsopp, J. F. 50, 113, 115 Bantelmann, J. 75
Alpert, N. M. 131 Barbrack, R. 372
Alsen, M. 155 Barker, W. W. 247
Amabile 422 Barkley, R. A. 198
Amaral, P. 201 Barlow, N. 418
Amato, D. A. 247 Barnes, G. 393
Amelang, M. 101, 194, 199 Barnes, J. 174
American Psychiatric Association 365, Barnes, T. R. E. 377
366 Baron, A. 478
Ammann, R. 377 Baron, M. 375
Anderson, B. 248 Barratt, E. S. 195, 196, 198, 199, 343, 344,
Anderson, J. E. 318, 319 358, 395
Anderson, K. J. 190, 202, 203 Barrett, G. 469
Angleitner, A. 73, 75, 79, 82, 198 Barrett, P. 77, 78, 79, 102, 194, 242, 245,
Aniskiewicz, A. 125, 126 300, 379, 462, 484, 485
Anokhin, A. 244 Barrett, P. T. 171, 172, 248
Ansbacher, H. L. 23 Barren, F. 410, 411, 435
592 Author index

Bartol, C. R. 124 Blair, H. 518

Bartram, D. 470, 471 Bleuler, M. 407
Bates, J. E. 73 Blinkhorn, S. F. 245, 293
Bates, T. C. 245, 302 Block, J. 5
Bayliss, G. C. 357 Bloom, J. R. 94, 97
Beach, R. C. 182 Blumenthal, T. D. 60, 394
Beane, W. E. 412 Blumstein, A. 134
Beauducel, A. 299 Blunt, P. 468
Beck, A. T. 374 Boas, F. 419
Beck, E. 245, 293 Boddy, J. 484
Beech, A. R. 357 Boggild, C. 444
Beh, H. C. 299 Boiarsky, G. 479
Bell, I. 394 Boissiere, M. 270
Belmaker, R. H. 8, 10, 63, 397 Boivin, M. J. 247
Bern, D. J. 534 Bond, N. 350
Ben-Shakhar, G. 516 Boomsma, D. I. 244, 248
Bendig, A. 465, 468 Booysen, A. 474
Benjamin, E. C. 502 Boring, E. G. 297
Benjamin, J. 8, 63 Bouchard, T. J. Jr 198, 241, 242, 243, 247,
Bennett, E. R. 8, 63 315, 424, 480
Benski, C. 492 Bouton, M. E. 354, 355
Bentall, R. 379 Bowen, M. 149
Bentler, P. M. 173, 174 Bowyer, P. A. 203
Benton, D. 103 Boyle, G. 379
Berardi, A. 131 Braden, J. P. 313
Berent, S. 247 Braden, W. 245, 293
Berger, T. W. 339 Bradley, A. 480
Berkeley, G. 570 Bradley, C. 149
Berlyne, D. E. 12, 199 Bradley, M. M. 56, 62
Berquist, A. 197 Brand, C. R. xvii, 23, 24, 25, 26, 29, 30,31,
Berrios, G. E. 374 32, 287, 301, 571
Best, S. 503 Brebner, J. 80, 358, 395
Bevill, M. J. 350 Brengelmann, J. C. 21, 516
Beyts, J. 344, 345 Brewer, W. F. 346, 350
Bhat, A. V. 116 Briggs, D. 518
Biferao, M. A. 351 Briscoe, D. M. 183
Bihrle, S. 132 Britt, T. W. 394
Binet 221, 285, 286, 291 Broadhurst, P. L. 199
Birbaumer, N. 349, 355, 434, 452 Brocke, B. 299
Birchall, P. M. A. 22, 71, 370 Brody, N. xxvi, 315, 573
Biron, L. 124 Brody, W. 267
Bishop, D. V. M. 5 Brooks-Gunn, J. 272
Bishop, D. 469 Broverman, D. M. 244
Bizot, E. 48 Brown, C. R. 168
Blackburn, R. 151, 152, 160 Brown, J. 302
Blackmore, S. 527, 531 Bruen, R. S. 375
Elaine, D. 8, 63 Brunas-Wagstaff, J. 197
 Author index 593

Buchsbaum, M. S. 10, 131, 132, 134, 247, Cattell, M. D. 20

397 Cattell, R. B. xviii, 20, 21, 22, 24, 25, 29,
Buchtel, H. 247 68, 72, 198, 219, 291, 411, 418, 422,
Buckalew, L. W. 395 434, 435
Buckley, P. 377 Ceci, S. J. 321
Bullen, J. G. 358 Cegavske, C. F. 339
Bullock, W. A. 393 Chadwick, S. B. 430
Burbeck, E. 472 Chaiken, S. R. 303
Burch, P. 92, 335 Chalke, F. 244
Burgess, P. K. 20 Chamberlain, A. G. 202
Burley, V. 104 Chambers, J. A. 411
Burns, J. 395 Chamove, A. S. 12
Burston, D. 364 Chang, J. Y. 247
Burt, C. L. xxiii, 18, 216, 219, 285, 301, 448 Chapman, J. G. 60
Buss, A. H. 6, 7, 72, 73, 74, 79, 195, 196, Chapman, J. P. 380
198, 371 Chapman, L. J. 380
Busse, T. V. 422 Chase, W. G. 290
Butcher, H. J. 434, 435 Checkley, S. A. 47, 60, 61, 394
Bylsma, F. W. 397 Chen, C. K. H. 505
Chess, S. 74
Caballo, V. E. 395 Chiang, A. 297
Cahan, S. 322 Chipuer, H. M. 241
Cameron, N. 408 Chizhevsky, A. L. 492, 493, 503
Campbell, A. 392 Chorney, K. 251, 252
Campbell, D. 501 Chorney, M. J. 251, 252
Campbell, F. A. 312, 325 Christal, R. E. 304
Campbell, J. B. 194, 200, 392, 394, 478 Christie, M. 469
Campbell, K. B. 245, 388 Christjohn, R. D. 151
Canter, S. 424 Christopher, M. 531
Capron, C. 326 Chueh, D. 247
Cardon, L. R. 243, 249 Ciaranello, K. 10
Carlson, J. S. 216, 299 Ciaranello, R. 397
Carpenter, P. A. 284 Cicchetti, D. V. 513
Carr, T. S. 182 Claridge, E. 409
Carrigan, P. M. 191 Claridge, G. S. xxvii, 22, 71, 357, 365, 368,
Carrillo-de-la-Pena, M. T. 195, 199 368, 370, 371, 372, 373, 375, 378, 379,
Carroll, J. B. 284, 289, 290, 291 380, 388, 430, 464, 573
Carver, A. 484 Clark, G. A. 348
Carver, C. S. 39, 52 Clark, H. H. 290
Caryl, P. G. 245, 246, 293, 294, 302, 303 Clay, J. 364, 365
Casal, G. B. 395 Cloninger, C. R. 8, 37, 53, 56, 60, 197
Casey, D. E. 377 Cochrane, R. 151
Caspi, A. 198, 321 Cohen, D. 463
Castelnuovo-Tedesco, P. 518 Cohen, J. 134
Casto, S. D. 241, 242 Cohen, N. 322
Catlin, N. 175 Colet, A. V. 294
Cattell, J. 410, 422 Collins, A. C. 63
594 Author index

Colman, A. 471 Davidson, A. B. 58

Colohan, H. A. 377 Davidson, K. M. 161
Colombo, J. 320 Davidson, R. A. 9
Conley, J. J. 178 Davies, D. R. 202
Cook, E. W. HI. 56 Davis, H. 23
Cook, L. 58 Davis, T. L. 56
Cook, M. 465, 471 Dawson, M. E. 131, 351, 356
Cooper, C. J. 245, 293 De Burger, J. 147
Cooper, C. L. 93 De La Casa, L. G. 22
Cooper, C. 358, 395 De Waard, R. 478
Cooper, H. 520 de Graaf, W. 503
Cooper, R. 466, 467 Dean, G. A. xxviii, 511, 513, 514, 517, 519,
Corcoran, D. W. J. 20 522, 523, 524, 525, 526, 575
Corder, E. H. 251 Deary, I. J. xxvi, xxviii, 24, 30, 246, 284,
Corley, R. 63 285, 287, 289, 292, 293, 297, 300, 301,
Corr, P. J. xviii, 38, 43, 45, 46, 47, 53, 54, 302, 303, 573
55, 60, 61, 66, 201, 394 Decker, S. N. 243
Costa, P. T. Jr xviii, 7, 24, 26, 37, 83, 190, DeFries, J. C. 63, 241, 242, 248, 250, 316
485 Delva, N. J. 375
Coudron, D. 492 Dennis, W. 405
Court, J. H. 272 Denny, M. R. 287
Cox, C. M. 410, 422 Derryberry D. 63, 198
Cox, D. N. 147, 172, 333, 334, 392 Desmarais, L. B. 328
Cramer, D. 178 Detterman, D. K. 251, 252, 287
Crayton, J. W. 397 Diaz, A. 39, 43, 49, 50, 52, 201
Cressey, D. R. 152 Dichgans, J. 349, 355
Croake, J. W. 175 Dickinson, A. 350, 354
Crombie, I. K. 104 Dickman, S. J. 196, 197
Cronbach, L. J. 290 Diener, E. 197
Crook, G. 73 Digman, J. M. 190
Crow, T. J. 375 Dodrill, K. L. 247
Cubos, P. F. 395 Dodson, J. D. 199
Cueto, E. G. 395 Doherty, N. 174
Curry, P. 526 Doiron, S. 377
Cuthbert, B. N. 56, 62 Donald, J. R. 370
Cutts, K. K. 149 Domic, S. 392
Doucet, C. 396
Dabbs, J. M. 182, 439 Douglas, J. 147
Daitzman, R. J. 181, 182 Douglas, V. I. 198
Dale, H. 470 Dreher, H. 95
Dambrot, F. 469 Drevdahl, J. E. 410, 411, 422
Daniels, J. 251, 252 Drwal, R. L. 77, 78, 79
Danmarks Statistik 434 Duara, R. 247
Danzinger, D. 568 Dubrieul, D. L. 393
Daum, I. 248, 349, 355 Dubrov, A. P. 503
Davey, G. C. L. 350 Dudek, S. Z. 435
David, A. 381 Duesberg, P. 419
 Author index 595

Duggan, C. F. 373 132, 133, 142, 143, 144, 145, 146, 148,
Dunbar, S. 475 149, 150, 151, 152, 153, 155, 158, 160,
Duncan, G. J. 272 165, 166, 167, 168, 169, 170, 171, 173,
Dundas, M. 116 174, 175, 176, 177, 178, 181, 183, 184,
Dustman, R. E. 245, 293 190, 191, 192, 193, 194, 196, 198, 199,
Duyme, M. 326 219, 220, 226, 228, 234, 235, 240, 241,
Dwyer, T. 511 242, 243, 245, 246, 248, 260, 261, 262,
282, 284, 285, 286, 287, 292, 300, 302,
Easterbrook, G. 275 303, 311, 312, 314, 317, 318, 320, 324,
Easting, G. 50, 115 325, 328, 333, 335, 336, 339, 340, 343,
Eaton, W. P. 202 344, 346, 350, 353, 354, 355, 358, 359,
Eaves, L. J. 112, 198 365, 366, 367, 368, 369, 370, 371, 372,
Ebstein, R. P. 8, 63 374, 375, 376, 379, 380, 381, 388, 391,
Eckblad, M. 380 392, 393, 394, 395, 404, 406, 407, 410,
Eddy, I. A. 503 412, 416, 417, 418, 419, 423, 424, 425,
Edelbrock, C. S. 129 426, 427, 428, 429, 430, 440, 441, 444,
Edell, W. S. 378 447, 448, 449, 450, 455, 456, 462, 463,
Edman, G. 195 464, 465, 466, 467, 470, 471, 474, 475,
Edwards, A. E. 165 476, 478, 479, 481, 485, 487, 491, 492,
Edwards, A. G. 517 493, 511, 512, 513, 514, 515, 516, 519,
Edwards, J. A. 9, 244, 397 520, 521, 522, 523, 524, 525, 526, 527,
Egan, V. 24, 30, 302 528, 529, 530, 531, 532, 533, 534, 536,
Eisinger, A. J. 173 537, 543, 544, 545, 547, 568, 569, 570,
Eisler, R. 497 574, 578, 583
Ekehammar, B. 200, 392 Eysenck, M. W. 5, 20, 28, 68, 71, 72, 73,
Ellingson, R. J. 244 78, 80, 82, 190, 393, 463
Ellis, H. 410, 417 Eysenck, S. B. G. xx, xxi, xxvii, 5, 6, 7, 21,
Ellis, L. 171, 182, 448, 452 26, 37, 50, 70, 77, 78, 79, 80,109,112,
Emmons, R. A. 197 114, 115,117,119,124,143,146,155,
Emran, A. 247 156, 160, 190, 191, 193, 194, 196, 342,
Endler, N. S. 405 368, 375, 391, 425, 520, 522
Engstrom, G. 155
Erasmus, J. 474 Faragher, E. B. 93
Errera, I. 495, 496, 498 Farkas, G. M. 168
Ertel, S. xxviii, 492,503,506,520,522,525, Farley, F. 395
574, 575 Farmer, A. 369, 381
Ertl, J. 244, 245 Farrington, D. P. 124, 154, 155, 156,
Estes, W. K. 287 160
Evans, I. M. 168 Feist, G. J. 412
Evans, R. B. 173 Feldman, J. 320
Ewing, J. A. 174 Feldman, M. P. 113, 115
Eysenck, H. J. xx, xxi, xxii, xxviii, xv, 3,4,5, Feldon, J. 409
6, 7, 11, 12, 17, 21, 22, 24, 25, 26, 28, Ferguson, L. W. 20
29, 32, 37, 60, 68, 70, 71, 72, 73, 76, Figlio, R. M. 160
77, 78, 79, 80, 82, 83, 92, 95, 96, 98, Filion, D. 131
99, 100, 101, 102, 104, 109, 110, 111, Fine, B. 473
112, 115, 117, 118, 119, 123, 124, 127, Fink, M. 375
596 Author index

Finn, S. E. 72 Garber, H. L. 325

Finner, S. 169 Garriger, M. S. 320
Fischer, B. 292 Gaskell, P. C. 251
Fisher, S. 173 Gatenby, S. 104
Fisherman, S. 271 Gates, A. I. 103
Flavell, R. 395 Gauquelin, F. 491,493,512,520,522,523,
Flew, A. 527 524, 525, 536
Flint, J. 63 Gauquelin, M. 491, 493, 512, 519, 520,
Florey, C. du V. 104 522, 523, 524, 525, 536
Flynn, J. R. 264, 266, 268, 275 Geddes, N. 104
Fohl, F. K. 502 Geen, R. G. 392, 393
Fonseca, A. C. 151 Geller, G. 58
Fotiadou, M. 47, 60, 61 Gentile, C. G. 348
Fox, B. H. 101, 103 George, E. I. 408
Frady, R. L. 182 Gerbing, D. W. 195, 1%, 198
Franks, C. M. 149, 368, 372 Getzels, J. W. 434
Fraser, I. C. 245 Giannitrapani, D. 244
Frcka, G. 194, 201, 344, 345 Gibb, C. M. 297
Frearson, W. M. 235, 242, 300 Gibson, G. J. 303
Freeman, A. 374 Gibson, H. B. 332, 333, 334, 336, 462, 511,
French, C. C. 531, 532 523, 544
French, J. R. Jr 417 Gibson, T. xvi, xxvii, xvi
Frency, C. xxviii Giese, H. 165
Frenzel, E. 498 Gilbert, B. O. 293
Frenzel, H. A. 498 Gilbert, D. G. 92, 293
Freud, S. xxviii, 29, 571 Gilliland, K. 190, 202, 393
Freudenthaler, H. H. 245 Gillispie, C. C. 497, 501
Friedman, A. F. 365 Gilmore, J. B. 527
Frier, B. M. 297, 302 Giordani, B. 247
Frith, C. D. 395 Glaser, R. 295
Fulford, K. W. M. 173 Glewwe, P. 270, 272
Fulker, D. W. 7, 63, 240, 241, 242, 250, Glover, J. A. 422, 423, 425
316, 529 Glymour, C. 513
Furedy, J. J. 351 Goertzal, M. 422
Furneaux, W. D. 284, 286 Goertzel, T. 422
Furnham, A. xxviii, 358, 376, 463, 472, Goertzel, V. 422
473, 475, 476, 477, 479, 480, 484, 486, Goldberg, L. R. 83, 197
516, 574 Goldman, S. R. 296
Fuzeau-Braesch, S. 522 Goma-i-Freixanet, M. 155, 200
Gooch, R. N. 370
Gabrys, J. B. 151, 153, 154 Goode, D. J. 397
Gahm, R. C. 395 Gorayska, B. E. 430
Gale, A. 9, 71, 199, 244 Gorman, L. 471
Galifret, Y. 492 Gorsuch, R. L. 21, 44
Galkowska, M. 175 Gosselin, C C. 171, 172, 173,
Gallon, F. xxiii, 215, 220, 238, 287, 297, 180
406, 410, 424 Gossop, M. 115
 Author index 597

Gottheil, E. 94, 97 Haier, J. A. 245

Gouchie, C. 439 Haier, R. J. 246, 247, 293
Gough, H. G. 157 Haines, J. L. 251
Gould, S. J. 284 Hall, C. S. 388
Graham, K. S. 297 Hall, W. 435
Granger, G. W. 21 Halverson, C. F. Jr 198
Grant, D. A. 352 Hamer, D. H. 63
Graves, P. 94 Hammer, D. H. 8
Gray, J. A. xiv, xvi, xxii, xiv, xvi, 10, 11, 12, Hammer, M. 430
22, 23, 36, 37, 38, 39, 43, 45, 46, 47, Hampson, E. 182, 438, 452
50, 51, 52, 53, 54, 55, 56, 60, 61, 62, Hampson, S. 267
63, 72, 75, 76, 112, 119, 151, 152, 180, Hanin, Y. 79
192, 201, 341, 344, 345, 349, 350, 355, Hansen, C. 474
369, 370, 371, 372, 373, 388, 389, 394, Hare, R. D. 117, 125, 126, 160
409, 478, 481, 482, 484, 571, 572 Hargrove, M. F. 182
Gray, N. S. xviii, 22, 23, 47, 60, 61, 63, 112, Harkness, A. R. 321
119, 409 Harlow, H. F. 12
Green, R. L. 322 Harper, P. S. 252
Greenberg, B. D. 8, 63, 479 Harrell, R. F. 103
Greene, J. 518 Harris, G. T. 377
Greenwald, A. G. 534 Harris, J. A. 182, 241, 248, 249
Greenwald, M. K. 56 Hassler, M. 434, 452
Greenwood, R. J. 47, 51, 52 Haupt, T. D. 166, 171
Greer, S. 94 Hawk, L. W. 56
Griffing, H. 300 Hawley, C. W. 392, 478
Griffith, M. 168 Haybittle, J. 94
Griffiths, P. D. 173 Haycock, J. 377
Gross, O. 69 Hayes, M. E. 322
Grossarth-Maticek, R. xx, 102, 104, Hazen, K. 247
335 Hazlett, E. 131
Grossberg, S. 40 Hebb, D. O. 4, 5
Gucker, D. K. 245 Hecker, R. 303
Gudjonsson, E. 471 Heffner, T. G. 398
Gudjonsson, G. H. xxi, 102, 123, 142, 145, Heller, J. F. 200
146,153,154,155, 157,158,160, 516, Helmers, K. F. 195, 198
572 Helmreich, R. L. 412
Guilford, J. P. 37, 72, 190, 192, 195, 197, Hemming, J. H. 125, 126
198, 422 Hemsley, D. R. 350, 358, 409
Gunter, B. 479, 516 Hendrickson, A. E. 245, 246, 283, 287,292
Gupta, A. 395 Hendrickson, D. E. 245, 246, 293
Gupta, B. 483 Hendrickson, E. 283, 287, 292
Gustafsson, J. E. 284 Hengevold, R. 503
Gustavsson, P. 155 Henri 286
Henry, C. E. 149
Haapasalo, J. 79, 80 Hepburn, D. A. 297
Hackley, S. A. 357 Herbison, G. P. 373
Hagopian, L. P. 43 Hernaiz-Sanders, H. 60
598 Author index

Hernandez-Peon, R. 357 lacono, W. G. 315

Hernstadt, R. 45 Iker, H. 93
Herrington, R. N. 370 Ingham, J. G. 369
Herrmann, W. M. 182 Insel, P. M. 174
Herrnstein, R. J. 269, 273, 416 Irving, K. 492, 525
Heston, L. L. 430
Heusel, C. 182 Jacklin, C. N. 434
Hewitt, J. K 529 Jackson, C. 485
Heymans, G. 69, 190 Jackson, D. N. 182, 197, 198, 249, 410,
Hichwa, R. 247 418, 422
Hick, W. 286 Jackson, M. 379
Hilgard, E. R. 350 Jackson, P. W. 434
Hill, A. 467 Jacob, F. 584
Hill,T. 48 Jacobs, G. A. 44
Ho, H. Z. 243, 244 Jacoby, H. 270, 272
Hodgson, R. J. 478, 479 Jang, K. L. 241, 248, 249
Hofmann, L. J. 322 Janke, W. 197
Hogan, T. P. 377 Janssen, R. H. C. 3%
Holland, P. C 351, 355 Jarrell, T. W. 348
Hollister, L. E. 375 Jarvik, L. F. 430
Holmes, R. M. 147 Jasper, H. H. 149
Honorton, C. 534 Jaynce, M. 20
Hopf, S. 173 Jean Renaud, G. 469
Horberman, R. B. 103 Jencks, C. 327
Horn, J. L. 245 Jensen, A. R. xvi, xxiii, xxiv, xvi, 21, 228,
Horn, J. M. 319 235, 237, 238, 242, 243, 248, 260, 261,
Horn, P. D. 395 262, 263, 265, 268, 274, 286, 297, 298,
Horvath, P. 200 299, 301, 302, 322, 417, 418, 572
Houlihan, M. 393, 395 Jessel, D. 156
Howarth, E. 151 Jessup, G. 470
Howells, K. 376 Jessup, H. 470
Hsu, C. C. 265, 267 Jimerson, D. C. 397
Hugdahl, K. 131, 339 Jin, Y. 263
Hull, C. L. 339, 341, 350 Jinks, J. L. 7
Hummel, H. 395 Joe, V. C. 168
Humphreys, L. G. 343, 350 John, O. P. 83, 197, 198
Humphreys, M. S. 190, 202, 203, 377 Johnson, S. xrii, 165, 293
Hung, S. S. 406, 448 Johnstone, E. C. 377
Hunt, E. 287, 289, 290, 296 Joireman, J. 6, 200
Hunter, J. E. 417 Jones, A. L. 370
Hunter, T. A. 297 Jones, D. M. 202
Huntly, B. J. P. 302 Jones, J. 343
Huntsman, R. G. 173 Jones, R. 168
Hurlbert, D. F. 174 Jones, S. H. 350
Husted, J. R. 168 Joseph, M. 73
Hutchings, B. 156 Josiassen, R. C. 245
Hyman, R. 286, 531, 535 Jouvet, M. 357
 Author index 599

Jung, C. G. xxviii, 368 Kneier, A. W. 94

Junginger, J. 377 Knight, J. B. 270
Just, M. A. 284 Knoop, A. A. 503
Koeppe, R. A. 247
Kagan, J. 73, 195, 198 Kohn, A. 533
Kamin, L. J. 273, 311, 324, 325, 311, 324, Kohn, P. M. 393
325 Kohnstamm, G. A. 198
Kane, F. J. 174 Kolb, D. 474
Kaniel, S. 271 Kolin, E. A. 5
Kanno, P. H. 174 Koopmans, L. 440
Kantor, L. 242 Korfine, L. 378
Kantorowitz, D. A. 180 Kornbrot, D. 535
Karlsson, J. L. 430 Kosslyn, S. M. 131
Kasarda, S. 251, 252 Kraemer H. C 94, 97
Kasmer, J. A. 194 Kraft, M. R. 6, 8, 171, 200
Kasparson, C. J. 422 Kranzler, J. H. 301, 302
Kastrup, K. W. 439, 453 Krebs, H. 197
Katz, M. 8, 63 Krech, D. 301
Kausler, D. H. 297 Kretschmer, E. 374, 450
Kaviani, H. 47, 60, 66, 394 Krippner, S. 527
Kayata, L. 352 Kroeber, A. L. 493, 494, 495
Kaye, M. 533 Kuhlman, D. M. 6, 200
Keller, J. F. 175 Kulhara, P. 375
Kelly, D. 20, 377 Kumari, V. xviii, 47, 60, 61, 66, 394
Kelly, E. L. 178 Kwapil, T. R. 380
Kelly, I. W. 525, 526 Kyllonen, P. C. 304
Kelly, J. 377
Kemp, E. 434 La Rue, D. 439
Kendel, E. 156 LaBuda, M. C. 316
Kendell, R. E. 375 LaCasse, L. 247
Kennedy, P. 174 Laing, R. D. 364, 365, 367
Kennedy, R. A. 104 Lally, M. 292, 301
Kerr, J. H. 200 Lane, D. A. 113, 151, 155
Kerr, W. A. 435 Lang, P. J. 56, 62
Keuss, P. J. G. 395 Langan, S. J. 297
Kim, J. 469 LaPierre, Y. D. 375
Kimmel, H. D. 350 Larsen, R. J. 202
Kimura, D. 438, 439, 452 Larson, G. E. 247, 302
Kinsey, xxii, 165 Lavond, D. G. 348
Kirkcaldy, B. 392, 395 Law, L. 478, 479
Kirkegaarde-Sorenson, L. 156 Laye, R. C. 151, 154
Kirton, M. 175, 468 Lazarus, A. A. 372
Kissen, D. M. 92, 93, 97 Le Moal, M. 398
Kitanawa, S. 265 Le Shan, L. 93
Klebanov, P. K. 272 Lea, S. 265
Klein, D. F. 375 LeBlanc, M. 124
Kline, P. 20, 294, 296, 462 LeDoux, J. E. 131
600 Author index

LeDoux, J. 349 Lucker, G. W. 265, 412

Lee, A. S. 373 Lucking, S. 300
Lees, J. L. 202 Luengo, M. A. 195
Lehrl, S. 292 Lukas, J. H. 10
Lehtinen, S. 247 Lunneborg, C. 287
Leith, G. 475 Lutzenberger, W. 349, 355
Lencz, T. 378 Lykken, D. T. 7, 127, 154, 180, 198, 243,
Lenzenweger, M. F. 378 315, 424
Lerner, R. M. 74, 79 Lynam, D. R. 149, 154, 160
Lessac, M. S. 147 Lynch, M. J. 203
Lester, D. 395 Lynn, R. riv, xvi, xxiv, xxv, xxviii, 171, 261,
Letemendia, F. J. J. 375 267, 270, 272, 273, 274, 275, 276, 572,
Levander, S. 377 573
Levey, A. B. 181, 200, 342, 343, 344, 345,
350, 352, 353 MacAndrew, C. 39
Levey, S. 376 Maccoby, E. E. 434
Levin, I. 477 MacKinnon, D. W. 422, 434, 435
Levoy, G. 101 Mackintosh, N. J. 271
Levy, P. 300, 303 Maclean, A. 171
Levy, S. 103 MacLeod, C. M. 289, 290, 372
Lewak, R. 178 MacLeod, K. J. 302
Lewicki, P. 48 MacMillan, D. 247
Lewis, D. 518 MacMillan, J. F. 377
Lewis, J. 287 Macy, R. 131
Lewontin, R. C. 273 Magnusson, D. 198
Li, D. 263 Magoun, H. W. 4
Li, L. 8, 63 Malmo, R. B. 390
Li, X. 264, 265 Maltzman, I. 125, 126
Light, R. J. 517 Maluish, A. M. 103
Lindsley, D. B. 4, 389 Mangan, G. 245
Lindsley, D. D. 149 Manning, A. A. 397
Lindzey, G. 333, 337, 388 Mansfield, P. S. 422
Lippman, M. 103 Mapperson, B. 303
Lipton, M. A. 174 Mares, M. 479
Livson, N. 301 Margittai, K. 397
Lloyd, J. E. M. 199 Marinkovic, K. 356
Loehlin, J. C. 319, 373 Markel, D. 247
Lombroso, C. 422 Martin, I. xxvii, xxviii, 20, 194, 201, 342,
Longstreth, L. E. 299 343, 344, 345, 350, 352, 353, 573
Loo, R. 200, 472 Martin, M. 373
Lord, J. 173 Martin, N. F. 10, 397
Lott, I. 247 Martin, N. G. 112, 179
Lottenburg, S. 138 Martin, R. P. 198
Lovibond, P. F. 350 Martin, T. 171
Lowenstein, D. A. 247 Martindale, C. 494
Lubow, R. E. 22 Marx, K. 570
Lucas, A. 178 Mascie-Taylor, C. G. N. 271
 Author index 601

Mason, O. 378, 379 Merkel, J. 286

Masters xxii, 165 Merry, J. 173
Mather, A. C. M. 523, 525, 526 Merskey, H. 368
Matthews, A. W. 372 Merwin, J. C. 264, 265
Matthews, G. 101, 199, 202 Messer, S. B. 198
Matthews, K. A. 412 Mey, J. L. 430
Mattie, H. 396 Meyberg, U. 168
Matussek, P. 408 Meyer, D. E. 196
Maxwell, S. E. 291 Meyer, F. 375
Maylor, E. A. 304 Michael, K. 9
Mayo, J. 520, 521, 522, 523, 529 Michaud-Achorn, A. 393
McCabe, P. 348 Mill, J. S. 570
McCall, R. B. 320 Miller, E. M. xxv, 277, 278
McCarthy, J. M. 241 Miller, L. T. 303
McClearn, G. E. 241, 248, 251, 252 Miller, S. 63
McClelland, D. C. 411, 435 Modestin, J. 377
McColloch, M. A. 293 Modgil, C. 216, 462, 511, 550
McCord, R. 483 Modgil, S. 216, 462, 511, 550
McCormick, D. A. 348 Moffitt, T. E. 198, 321
McCrae, R. R. xviii, 7, 24, 26, 37, 83, 190, Moir, A. 156
485 Monahan, J. 377
McCreery, C. 379, 380 Morgan, J. 47, 51, 52
McCrimmon, R. J. 302 Morgan, R. F. 322
McDougall, C. 154 Morgenstein, S. 478, 479
McDougall, W. 32 Mori, M. 247, 248
McEwan, A. W. 154 Morris, G. L. 294
McEwen, B. S. 442 Morris, J. 194
McFatter, R. M. 197 Morris, T. 94
McGarry-Roberts, P. A. 245, 395 Morrish, R. B. 392
McGue, M. 180, 243, 315, 424 Morrison, F. J. 322
McGuffin, P. 251, 252 Morse, R. J. 322
McGuire, R. J. 368 Moruzzi, G. 4
McGurk, B. J. 154, 260 Mosher, D. L. 158
McKenzie, J. 291, 478 Mowbray, R. M. 368
McLearn, G. E. 316 Mueller, E. F. 417
McLeod, H. 388 Mullen, P. E. 373
McMichael, A. J. 315 Mullholland, T. M. 295
McNeil, G. 104 Mulligan, G. 468
McNeil, T. F. 430 Munjack, D. J. 174
Mealey, L. 158, 159 Murphy, D. L. 8, 10, 63, 397
Medhurst, S. 476 Murphy, G. 3
Medicus, G. 173 Murray, C. 269, 273, 416
Mednick, S. A. 156, 127, 129, 378 Murray, H. A. 190, 191, 195
Meehl, P. E. 378 Murray, H. G. 412, 413, 416, 435
Meltzer, H. Y. 397 Murray, R. M. 373
Menzies, I. 104 Muse, K. B. 60, 394
Mercier, L. 397 Myslobodsky, M. S. 382
602 Author index

Naatanen, R. 357 Okaue, M. 470

Nador, S. 242 Ollendick, T. H. 43
Naera, R. 439, 453 Oniszczenko, W. 80
Naismith, D. J. 104 Oppenheimer, R. 418
Nakamura, M. 470 Oreland, L. 397
Narrol, R. 502, 503 Organ, D. 468, 476
Neale, M. C. 243, 249 Orlebeke, J. F. 395
Nebylitsyin, V. D. 341, 370 Osgood, C. E. 158
Neisser, U. 273, 274, 286 Osher, Y. 8, 63
Nelson, J. H. 524 Otero, J. M. 195
Nelson, M. 104 Overall, J. E. 375
Nelson, R. D. 534 Owen, K. 269, 270
Nemanov, L. 8, 63 Owen, M. J. 251, 252, 381
Nesselroade, J. R. 316 Oziel, L. J. 174
Nestor, P. G. 377
Neter, E. 516 Palmer, J. 527, 536
Nettelbeck, T. 245, 292, 293, 300, 301, 302 Paltiel, L. 462
Netter, P. 398 Parker, J. D. A. 196, 197
Neubauer, A. C. 200, 245 Parks, R. W. 247
Newcomb, M. D. 174 Parsons, L. B. 127
Newman, J. P. 44, 45, 87, 195, 201, 484 Paspalanova, E. 272
Nias, D. K. B. xx, xxviii, 174,392,492,511, Passingham, P. E. 20, 124, 144, 151
513, 523, 524, 525, 575 Patrick, C. J. 56, 62
Nichols, R. C. 424 Patterson, C. 8, 63
Nicholson, J. 395 Patterson, G. R. 147
Nielsen, J. 439, 453 Patterson, M. M. 339, 484
Nieschlag, E. 434, 452 Patton, J. H. 195, 196, 198
Niura, K. 470 Paunonen, S. V. 412, 413, 416, 435
Nottebohm, F. 441 Pavlides, C. 442
Novick, M. 475 Pavlov, I. P. 69, 75, 76, 339, 340, 341, 347,
Novick, O. 8, 63 353, 354
Nuechterlein, K. 131 Payne, R. W. 408, 466, 467
Nyborg, H. xxv, xxvii, 181, 183, 423, 432, Peak, H. 297
433, 434, 436, 437, 438, 439, 440, 441, Pearson, P. R. 50, 115
444, 446, 448, 449, 452, 453, 455, 567, Pearson, R. xv, xvii
568, 569, 574, 577, 578, 580, 581, 582 Pearson, T. 94
Pedersen, N. L. 316
O'Callaghan, E. 377 Pedler, K. 535
O'Connor, K. 396 Peeler, W. H. 173
O'Gorman, J. G. 9, 199 Pelf, D. C. 418
O'Haire, T. 151, 154 Pellegrino, J. W. 295, 296
O'Sullivan, D. M. 171 Pellett, O. 245
Ochse, R. 422 Pelosi, A. J. 99
Oden, M. H. 422, 435 Pelosi, A. 381
Ogden, M. E. 116 Penney, J. 247
Ohlin, L. E. 155, 160 Pericak-Vance, M. A. 251
Ohman, A. 351, 352, 353 Perlstein, W. M. 9
 Author index 603

Perone, M. 478 Quay, H. C. 127

Perris, C. 10
Persson, G. 182 Rabbitt, P. M. A. 304
Pervin, L. A. 68 Rabin, A. 518
Peters, K. 151, 154 Rachman, M. A. 114, 156
Petersen, R. S. xxviii Rachman, S. J. 372
Peterson, E. 479 Radford, J. 422
Peterson, G. W. 178 Radin, D. I. 534
Petrill, S. A. 251, 252 Raine, A. xxi, 125,126,127,128,129,130,
Pettingale, K. 94 131, 132,133, 135, 136, 138, 151, 155,
Petursson, H. 157 156, 378, 380
Pfrang, H. 168 Ramey, C. T. 312, 325
Pfurtscheller, G. 245 Rammsayer, T. 398
Phillips, K. 316 Randi, J. 532, 533
Phillips, N. 151, 154 Rao, K. R. 527, 528, 536
Pickering, A. D. xviii, 22, 23, 38, 39, 43, 45, Rasmussen, K. 377
46, 49, 50, 52, 53, 54, 55, 60, 63, 112, Rausche, A. 473
119, 201, 409 Raven, J. C. 272
Piera, P. J. F. 294 Rawlings, D. 196
Pihl, R. O. 195, 198 Rawlins, J. P. 409
Pillemer, D. B. 517 Reber, A. S. 45
Pinneau, S. R. 318 Reed, M. A. 63
Pivik, R. T. 397 Reed, T. E. 238, 247, 248
Plomin R. 6, 7, 8, 73, 74, 79,195,196,198, Rees, J. 376
241, 242, 248, 250, 251, 252, 316, 371 Rehberg, R. A. 327, 328
Plooij-Van Gorsel, P. C. 396 Reinhardt, R. 470
Plum, A. 245 Reinisch, J. M. 183
Polani, P. 173 Rescorla, R. A. 339, 346, 350, 351, 356
Polkey, C. E. 349, 355 Resnick, L. B. 287
Popper, K. 520 Revelle, W. xxii, xxiii, 39, 43, 63, 190, 194,
Popplewell, D. 379 195, 198, 201, 202, 203, 572
Post, R. M. 397 Reynolds, C. R. 263, 422, 423, 425
Powell, G. E. 23, 151 Reynolds, G. 131
Powell, J. H. 47, 51, 52 Reynolds, J. H. 194
Power, M. 73 Rhodes, L. E. 245, 293
Price, D. 448 Riad, J. K. 182
Price, L. 5 Rice, M. E. 377
Priel, B. 8, 63 Rich, S. 198
Prior, M. R. 73 Richard, L. S. 178
Pritchard-Levy, A. 299 Rigby, R. G. 168
Przybecky, T. R. 37 Rijsdijk, F. V. 244, 248
Puckett, J. M. 297 Riley, D. M. 351
Pueyo, A. A. 294 Rim, Y. 465, 467, 468
Purifoy, F. 440 Roberts, G. 103
Putnam, H. 584 Roberts, J. C. 157, 158
Roberts, R. D. 299
Quadagno, D. M. 183 Roberts, R. J. 315
604 Author index

Robertson, E. F. 315 Ryle, G. 576

Robertson, G. 151, 154
Robins, R. W. 198 Sabot, R. H. 270
Robinson, D. L. 22, 71, 245, 293 Saccuzzo, D. P. 302
Robinson, J. O. 369 Sackett, P. R. 328
Robinson, T. N. 395 Saipe, J. 473
Rocklin, T. 39, 190, 194 Salters, A. 147
Rodnight, E. 370 Sandman, C. 247
Roe, A. 410, 422, 430, 431, 432, 434, 435, Sandra, E. A. 20
437, 438, 442, 443, 454 Sano, H. 264, 265
Roediger, H. L. 405 Sargent, C. L. 335,519,529,530,531,532,
Roemer, R. A. 245, 339 533, 534, 535
Roger, D. 194 Sargent, W. 287
Romans-Clarkson, S. E. 373 Saunders, A. M. 251
Romero, E. 195 Savage, R. 467
Ronai, E. 448 Savory, M. 202
Ronning, R. R. 422, 423, 425 Saxton, P. M. 10
Rose, R. M. 182 Scarpa, A. 129
Rose, S. A. 273, 320 Scarr, S. 273
Rosenburg, R. 156 Schafer, E. W. P. 244, 245, 246
Rosenhan, D. L. 430 Schalling, D. 155, 157, 190, 195, 196, 198
Rosenthal, E. R. 327, 328 Schell, A. M. 131, 351, 356
Rosenthal, R. 514, 534 Schenk, J. 168
Roses, A. D. 251 Schenk, J. 473
Rosolack, T. K. 83 Scherer, H. 357
Ross, E. 370 Scherg, H. 101
Ross, J. 191, 480 Schif&nann, K. 351
Roth, E. 286, 292 Schimdt-Rathjens, C. 101
Roth, J. 134 Schlien, B. 103
Rothbart, M. K. 198 Schmale, A. H. 93
Rovai, E. 406 Schmechel, D. E. 251
Rovine, M. 241 Schmeidler, G. R. 533
Rowley, K. 484 Schmidt, A. 165
Roy, A. 368 Schmidt, G. 168
Royer, F. L. 287, 288 Schneiderman, N. 348
Rubin, D. B. 514 Schoenhaler, S. 103
Ruch, W. 73, 75, 79, 82 Schofield, M. 165
Ruiz, G. 22 Schonell, F. J. 25
Rusalov, V. M. xix, 76, 79, 80 Schooler, C. 10
Rushton, J. P. xxv, 151, 178, 182, 274, 278, Schuessler, K. F. 152
392, 405, 409, 410, 412, 413, 416,419, Schugens, M. M. 349, 355
422, 435, 574 Schulsinger, F. 156
Rushton, P. xxvii Schumph, D. 151, 154, 154,
Russell, B. 513 Scott, R. 266
Russell, R. J. H. 178 Seagraves, R. 116
Rust, J. 155 Sears, D. J. 147
Rusted, J. M. 202 Segal, N. L. 198, 480
 Author index 605

Seifter, J. 58 Smith, M. 535

Seligman, M. E. P. 352, 430 Smith, P. L. 291, 301
Sellin, T. 160 Smith, S. L. 388, 392
Shaffer, J. 94 Smithers, A. G. 520
Shagrass, C. 245, 370 Smithson, P. 357
Shamberg, N. 395 Snow, R. E. 288
Shanks, D. R. 45, 350 Solomon, A. 103
Shannon, S. 349, 355 Solomon, R. L. 147
Shapiro, S. 442 Sommerlund, B. 580
Shaw, L. 472 Son, M. J. 267
Shea, J. D. 168 Sonnemann, U. 518
Sheard, C. 131 Sorokin, P. 493
Shell, P. 284 Sosa, E. 247
Sheramata, W. A. 247 Sparrow, N. H. 191
Shields, J. 18 Spearman, C. xxiii, 216, 219, 220,235, 238,
Shockley, W. xvi, 405 284, 285, 286, 287, 291, 303, 317
Shope, D. F. 173 Spence, J. T. 342, 346, 350, 412
Shucard, D. W. 245 Spiegel, D. 94, 97
Shuey, A. M. 260, 261 Spielberger, C. D. 44
Sichel, H. 472 Squire, L. R. 348, 349
Siddle, D. A. T. 127, 350, 392 St. John, M. F. 45
Siegel, J. 10 Stabholz, M. S. 516
Sigal, J. J. 368 Stahl, S. M. 397
Sigurdardottir, H. 157 Stanford, M. S. 358
Sigusch, V. 168 Stanley, J. 138
Silva, P. A. 198, 321 Star, K. H. 368
Simon, T. 221 Starosta-Rubinstein, S. 247
Simon, H. 398 Startup, M. 522
Simon, L. 202 Stavidrou, A. 358
Simonton, D. K. 404, 494, 495, 496, 501, Staw, B. 480
502, 503, 507, 544 Steadman, H. 377
Simpson, E. L. 430 Steele, T. 39
Sine, L. F. 168 Stein, L. 10
Singh, J. 350 Stelmack, R. M. xxvii, 9, 71, 245, 388, 390,
Singh, K. K. 156 392, 393, 394, 395, 396, 397, 574
Sinha, S. N. 274, 417 Stenberg, G. 199
Skuder, P. 251, 252 Stephenson, G. M. 20
Skulason, S. 157 Stern, L. 469
Slade, P. 368, 379 Sternberg, R. J. 284, 287, 294, 295, 422
Sletvold, H. 377 Sternberg, S. 297
Small, G. W. 251 Stetinsky, D. 272
Smith, A. D. 409 Stevenson, H. W. 265, 313
Smith, A. P. 202 Stevenson, V. E. 56
Smith, B. D. 9 Stewart, R. A. 23, 467
Smith, C. W. 503 Stigler, J. W. 265, 313
Smith, D. L. 251, 252 Stoddard, J. 132, 138
Smith, L. 322 Stokes, J. 477
606 Author index

Stomgren, E. 375 Thornquist, M. H. 6, 358

Stough, C. K. K. 80, 245, 292, 293, 302 Thornton, J. C. xviii, 44,46, 47,59, 60,61,
Straumanis, J. J. 245 394
Strelau, J. xix, xxviii, 69, 72, 73, 75, 79, 80, Thouless, R. H. 513
82, 83, 198, 571 Thurstone, L. L. 20, 216, 285
Strieker, E. M. 398 Titchener 570, 575
Stripp, A. 73 Tobena, A. 39
Strittmatter, W. J. 251 Tobera, A. 484
Strykowski, B. F. 406, 448 Todman, J. 104, 297
Suci, G. J. 158 Tomassen, G. J. M. 503
Sulloway, F. J. 410 Tomaszewski, T. 76
Svanborg, A. 182 Torrance, E. P. 434
Svebak, S. 200 Torrubia, R. 39, 484
Svrakic, D. M. 37 Touchette, P. 247
Swanck, R. 94 Traskman-Bendz, L. 155
Swanson, J. 377 Trasler, G. B. 127, 151, 154
Swartzentruber, D. 355 Travis, E. L. 297
Symonds, P. 544, 545, 546 Trouton, S. S. 149
Syndulko, K. 125, 126 Tunstall, O. 467
Szelenberger, W. 393 Turkheimer, E. 326
Turner, L. H. 147
Taft, R. 395 Turriff, S. 201
Takeuchi, M. 266 Tushup, R. 169
Tang, C. 263 Tyrer, P. 161
Tannenbaum, P. H. 158 Tyson, G. A. 522
Tanner, J. M. 173 Tzenova, B. 272
Taussig, H. N. 62
Taylor, C. B. 94 Ullwer, U. 194
Taylor, C. W. 410, 411, 435 Umansky, R. 8, 63
Taylor, M. A. 375 Utendale, K. A. 151, 154
Taylor, P. J. 377 Utts, J. 513, 514, 534, 535
Teasdale, J. 115
Teich, A. 348 Vagg, P. R. 44
Tellegen, A. 197, 198, 424 Vallance, R. C. 368
Temoshok, L. 94, 95 Van der Werff, J. J. 190
Teplov, B. M. 76 Van Kampen, D. 23
Terman, L. M. 410, 411, 422, 435 Van Zelst, R. H. 435
Teta, P. 6, 200 Vandenberg, S. G. 263
Teylor, T. J. 339 VanDercar, D. H. 348
Thalbourne, M. A. 533 Venables, P. H. 125, 126, 127, 128, 129,
Tharp, V. K. 125, 126 130,131,132,133,135,136,151,155,
Thomas, A. 74 156, 469, 473
Thomas, D. R. 168 Venditti, C. P. 251
Thompson, L. A. 251, 252 Vermolayever-Tomina, L. B. 478
Thompson, R. F. 285, 339, 347, 348 Vernon, P. A. xxiv, 241, 242, 243, 244, 247,
Thompson, W. L. 131 248, 249, 299, 303, 304
Thome, F. C. 166, 171 Vernon, P. E. 268, 407, 416, 515
 Author index 607

Vernon, P. 422, 572 White, T. L. 39, 52

Verster, J. 190 Whitehouse, M. 173
Vickers, D. 291, 300, 301 Whitely, S. E. 296
Vignetti, S. 251, 252 Wickett, J. C. 247
Vimpani, G. V. 315 Widaman, K. F. 216, 299
Vincent, K. R. 273 Wigg, N. R. 315
Vining, D. 434 Wilcox, K. J. 198
Visher, C. 134 Wilk, S. L. 328
Vogel, W. H. 244, 398 Willerman, L. 319
Vogel-Sprott, M. D. 59 Williams, D. 245, 293
von Knorring, L. 10, 397 Williams, J. M. G. 372, 373
Williams, L. 378
Wachs, T. D. 73 Williams, M. 132, 133, 135, 136, 151, 155,
Wadsworth, S. J. 324 156
Wagner, A. R. 339, 346, 350 Williams, P. M. 439
Wagstaff, G. F. 197 Williams, T. I. 497, 501
Wahlsten, D. 444 Willson, R. J. 300, 301
Wakefield, J. A. 174, 175, 176, 177, 178, Wilson, G. D. xxii, xxviii, 23,47, 60, 61, 66,
365, 483 147,169,170,171,172,173,175,180,
Walberg, H. 406, 448 183, 394, 467, 484, 485
Waldman, I. D. 273 Wilson, J. Q. 155, 160
Walkenfeld, F. F. 45 Wilson, K. G. 393
Walker, C. E. 168, 530 Wilson, R. S. 315, 319
Wallace J. F. 44, 45, 201 Winblad, B. 397
Wallach, M. A. 395 Windle, M. 73, 74, 79, 80
Walshe, D. 377 Wiseman, R. 535
Walton, V. A. 373 Wissler, C. 297
Wankowski, J. 467 Witkin, H.
Warburton, D. M. 397 Wolfe, R. N. 194
Wardell, D. 20 Wolfgang, M. E. 160
Wardle, J. 104 Wolfson, R. 515
Watts, F. N. 372 Woody, E. 409
Webb, E. 285 Woodyard, E. 103
Webster, C. D. 196, 197 Wrenn, R. L. 430
Weinberg, H. 245 Wyatt, R. J. 10, 397
Weinberg, R. A. 273
Weinstock, C. 101 Yaffe, M. 166
Weiss, V. 252, 292 Yerkes, R. M. 199, 260
Wells, P. A. 178 Yoshii, F. 247
Werre, P. F. 396 Young, A. 247
Westlind-Danielsson, A. I. 442 Young, R. A. 339
Weyers, P. 197 Young, S. N. 195, 198
Whang, P. A. 268 Yule, W. 151, 153
White, J. H. 20
White, K. D. 392 Zacherl, M. 477
White, M. 303 Zacune, J. 116
White, O. 520 Zahn, T. P. 10, 395
608 Author index

Zaleski, Z. 175 Zinser, M. C. 380

Zawadzki, B. xix, 73,76,79,80,82,83,571 Ziskind, E. 125, 126
Zeaman, D. 288 Zoob, I. 5
Zentall, S. S. 198 Zubek, J. 5
Zentall, T. R. 198 Zuber, I. 200
Zhang, Y. 293 Zubin, J. 430
Zhu, Y. 263 Zuckerman, M. xvii, xxviii, xix, xvii 5, 6, 7,
Ziegenfu, W. 497 8, 9, 10, 11, 37, 49, 63, 73, 74, 75, 79,
Zigman, M. J. 398 82, 112, 151, 153, 165, 169, 171, 182,
Zimmerman, W. S. 37, 195, 197 190, 195, 198, 200, 358, 371, 376, 397,
Zinbarg, R. 43, 63, 201 422, 449
Zindi, F. 270, 271
Subject index

Abecedarian Project xxvi, xxvii, 312-13, Mayo sun-sign study 520-2, 523
324, 325 planet-birth correlations 491, 492
accidents research 523
personality and 472-4 status of conclusions 525-6
addiction 115-16 survey of scientific evidence 523-5
alcoholism 116 augmenting 10
Alzheimer disease
specific gene 251 behavior therapy 353-4, 356, 462
studies of 246-7 emotional factors 354
anxiety 153, 371, 573 fear conditioning and extinction 354-5,
behavioral inhibition system (BIS) 153, 356
193 movement 372
classical conditioning and 342 theory 354
impulsivity and 203 behavior-genetics 462, 480, 572
sensitivity 201 adoption studies 323
social 60 creativity 407
Taylor Manifest Anxiety Scale 342 EAS Temperament Survey (EAS-TS)
arousal xvii, 20, 38, 146, 156, 371, 573 79, 80, 81
ascending reticular activating system effects of schooling 323-4
(ARAS) 389, 391, 394 heritability 243, 249-51
classical conditioning and 133, 134, 152, intelligence studies 240-4
156, 342, 356, 467 quantitative trait loci (QTL) xxiv,
cortical responses 9-10, 340, 398 251-2, 253
extraversion and 9, 199, 201, 342, 343, theory of temperament 73-4
388-90, 398, 399 behavioral activation system (BAS) xxii,
impulsivity 190, 198-203, 573 38, 39, 40, 41, 42, 49, 52, 56, 57, 58,
optimal level of 4, 5, 8 60, 153, 193
role of in personality xvii, 8-10 link to fight-flight system 39
theory of E and N dimensions 5 behavioral inhibition system (BIS) 38, 39,
underarousal and antisocial behavior 133 40, 49, 53, 56, 57, 58, 60, 153, 193
work mood studies 469, 478 link to fight-flight system 39
astrology 492, 511, 519-26, 534, 574 brain 355
Committee for Objective Research in creativity and neural plasticity
Astrology (CORA) hypothesis 440-1, 442, 449, 451
Eysenck's involvement 519-26 glucose metabolic rate (GMR) 225,238,
Fuzeau-Braesch twin study 522 244, 246-7
Gauquelin, M. and F. 335, 491, 492, imaging xvii, 13, 131-3, 137, 138
519, 520, 525, 536 magnetic resonance imaging (MRI) 225
610 Subject index

brain (continued) orienting, effect of 131, 134, 136, 156

nerve conduction velocity (NCV) 225, performance studies of 343
238, 244, 247-8 phobias 352-3, 357
neurophysiological research 347-9 quasi-conditioning 125, 126
racial differences in size of 262, 272, Rescorla-Wagner theory 350
273-4, 279 reward and punishment hypothesis 345
Burch, P. 335 role of xxviii, 17, 122, 123
Hurt, C. 216, 217, 218, 219, 545, 547, 556 socialization and 123, 124, 148
Squire's theory 349
cancer 92, 93-5, 572 stimulus conditions 343
cancer-prone personality traits 95, 96, studies of 127-8
101, 102 theory 339
psychological intervention 94, 97-8 Cloninger, C. R.
smoking and, 96-7 Harm Avoidence scale 53, 56, 60
spontaneous remission 101-2 Reward Dependence scale 53
role of immune system 103 cognitive performance
Carver, G. W. 311-12, 328, 410 cognitive ability 297-300, 406
Cattell, R. B. xviii, 218, 219 cognitive-behavioral debate 349-53
inspection time investigation 300-1 cognitive components 294-6, 354
children decision tasks and extraversion 395-6
extraversion 112 decision tasks and introversion 395-6
stimulant drugs, effect on 149 electrodermal conditioning 351, 352,
classical conditioning 151, 160 353
animal conditioning theory 346-7, 355 human eyelid performance 351-2
antisocial behavior 123, 125, 126, 127, impulsivity 201
128, 129, 131, 133, 143, 144 inspection time 301
anxiety 342, 345 memory trace system 348-9
arousal 133, 134, 152, 156, 342, 356,467 phobia studies 352-3
assessment of studies 126-7 psychopathology, cognitive dimension
associative theory 340 372, 382
brain circuits 340 television and music, effect of 479, 480
brain imaging and 131-3, 137, 138 use of term 351
cognitive-behavioral debate 340, conscience, 142, 144
349-53 components of 148
conscience and 144 concept of 123, 124
criminals and 125 development of 146-8
decline in interest 339 Costa and McCrae
emotions 339-40 big-5 model xviii, 7, 23, 24, 25, 31, 37,
Eysenck, biosocial theory of 127-9, 110, 197
129-31, 340, 341-6, 353, 355, 358, 359 creativity 404, 406
future research trends 355-9 arousal and motivation 417
human eyelid conditioning studies birds 441
343-5 children 441-2, 451-2
impulsivity and 200-1, 343-4 covariant trait development 436-44
Mauritius longitudinal study 129-31 domain specific 434
neurophysiological research 347-9 dopamine levels 404, 409, 426, 429, 449,
neuroticism and 342 574
 Subject index 611

emergenesis 424 regular succession 498

Eysenck's theory of 406-10, 412, 416- simultaneous and culturally
17, 419, 425, 428-30, 447, 455, 574 independent 498-500
females 439-40, 452 solar activity 503-7
General Trait Covariance Model crime 22
(GTC) 436-7, 439, 442, 444, 449, 455 classical conditioning, 125
genetic basis 424, 428-9, 430 Eysenck, theories of xxi, 123-4, 127-9,
genius 404, 417-18, 419, 423-4, 428, 129-31, 133, 142-54, 151-4, 160, 572
443, 450, 451, 452, 453, 455, 574 genetic basis 124, 145
hormotypic comparisons 437-8 hormone levels 182
importance of 422 personality, relationship with xxi, 143,147
index of 415, 416 prevention of, theory 143-4
latent inhibition 409 protective factor 133-7, 145
molecular 422-54, 456 psychoticism 138
molecular model (ND4M) for 444-5, sex and xxii
446, 447, 449, 450, 455 social causes xxi, 146
natural science model 430 crime, treatment of xxi
neural plasticity hypothesis 440-1, 442, drugs 149
449, 451 criminal behavior 20, 26, 129, 131-3, 137,
overinclusive thinking 408-9 144, 376
personality 406, 410-12, 435, 437 augmenting 10
productivity 406 aversion therapy 150
professors, studies of 412-16 big-6 (Cattell) xviii, 21, 28, 30
psychoticism 112, 117, 404, 407-8, 409- comparison of 22
10, 413, 415, 416, 429, 449-51, 574 description of elementary traits 145-6
research, future of 422-5, 454-6 effect of IQ 29
Roe, A., eminent scientists study 431-4, effect of maturation 27
454 Eysenck's theory of PEN 73, 75
schizophrenia 407, 408-9, 427, 429, 430, link to dimensions xxi, 142, 155, 345
450, 451 personality traits 155
scientific achievement 404—5 predictors of 156
IQ 407, 409, 416-17, 428, 434-5, prevention and treatment, Eysenck's
436-44, 574 theories for 148-50, 161
sociability 434-5 protective factors 145, 155-6, 160
socialization theory 424, 449
trait 427 dimensions, personality xvii, xviii, xix, xxi
creativity bursts and sunspot cycles xxviii, 17-20, 28-9, 37, 71, 152, 154, 379,
491-506, 574 472, 485
constant probability of success 495, 501 E dimension, see separate entry
co-occurrence 498 elementary traits, description of 145-6
cultural florescence 502, 506 human eyelid conditioning 343-5
emulation 494, 503 N dimension, see separate entry
independent of number of P dimension, see separate entry
creators 501-3 PEN, description of 70-1, 75
Kroeber, A. L., cultural evolution sex and, xxii
493-4, 507 temperament, relationship to xix, 71-3
production or creation 495-8 variation in number 25-7, 28
612 Subject index

E dimension 3, 9, 17-20, 60-1, 69, 152, Eysenck, H. xiv-xvii, xxi, xxii, xxvi, xxvii
154, 155, 160, 168, 178, 197, 379, behavior therapy 353-5
380, 571 classical conditioning theory, 127-9,
masculinity and 169, 170 129-31, 340, 341-6, 353, 355, 358,
sex and 26, 171 359
existence diverse interests 332-8, 462
molecular model (ND4M) for 444-5, impulsivity, work on 190
446, 447, 449, 450, 451-3, 454, 455 Pavlov xviii, 4, 36, 37
extraversion xxvii, 8,17,20, 26, 28, 68, 70, rebel, as a 334-5, 36
71, 72, 74, 87,109,119,124,142,143, sexual behavior research 165-84
151,152,154,159,160,182,370,390- sport and 333-4
2, 573, 574 Eysenck and creativity 404, 456
accidents and 472-3 approach to 425-6, 428
arousal and 9, 199, 201, 342, 343, model 426-8, 428-30, 447, 455
388-99, 399 theory of 406-10, 412, 416-17,419,425,
arousal theory 388-90 428-30, 447, 455, 574
astrology 520 Eysenck and crime 133, 137-8, 142-54,
children and 112 159-61
components of 193 biosocial theory of xxi, 127-9, 572
conditioning and 147,153,343,353,358 conditioning theory of 123-4
concientiousness versus, 26-7 criticism of theories 151-4, 160
description of 165, 167 prevention and treatment, theories for
dopamine activity (DA) 397-8, 399 148-50, 161
drugs, effect on 4, 149, 370-1 psychoticism 138
E dimension see separate entry Eysenck and fringe sciences xv, xxviii,
Eysenck's theory of 69 511-13, 536, 537
impulsivity xxii, 190,191,192-3,197, 201 astrology 335, 462, 492, 511, 519-26, 574
motor expression 394-6, 398 graphology 511, 514-19, 574
military pilots 470, 471 heliobiology 492, 493, 503, 507
neuroticism, effect of 470-1 parapsychology 492, 511, 526, 529-30,
organizations 484, 485 532, 574
reward and punishment 481-2, 483-5 Eysenck and intelligence 214-20,217,226,
television and music, effect of 479, 228, 234, 235, 242, 252-3, 282-7, 291,
480 292, 293, 302, 303, 328
teaching methods 475, 476 classification of phenomena 283
scale 192 conceptual aspects of 314
sensory psychophysiological evidence Galton-Spearman-Eysenck school 220,
393^, 573 221, 224-6
sexual behavior xxii, 167, 173-4, 179, inheritability of 240-2
180, 181, 183, 184 International Society for the Study of
stimulation sensitivity 388, 390, 392-4, Individual Differences 292
398, 573 models 248
traits of 192, 391-2 race and 259-61, 278
underarousal and 133 research 219, 283-7
vocational preferences 465, 466-7, 468, speed-IQ, biological approach 243
469 Eysenck and medical (health) psychology
work environment 472, 478-9 92-100
 Subject index 613

Grossarth-Maticek, R., collaboration factor analysis see also Spearman

with xx, 92, 93, 95-8, 101, 335 common factor variance 224
Eysenck and organizational psychology Freud 19, 21, 26, 27, 31, 32
462-87, 574 Furneaux, D.
early studies of 465-70 IQ and mental speed 284, 285-6
selection 462-3 Nufferno Test 217-8
personality profiler (EPP) 485, 574
work psychology 463 Gallon, F. 215, 216, 219, 220, 556, 568
Eysenck and personality 395, 398 genius 424
arousal theory of extraversion 388, 391, reaction time (RT) 219, 226, 227, 228,
394, 398 229
biological basis 71 Gauquelin, M. and F. 335, 491, 492, 519,
evolution of 3-5 520, 525, 536
Extraversion and Neuroticism scales 38, General Trait Covariance (GTC) model
68-70, 371 436-7, 439, 442, 444, 449
gigantic-3: xviii, 24 genetics
London School, and xxiii, 25, 214-15, crime 124, 145
216, 219, 485 personality and 7-8
P dimension, development of xx, 21 sexual behavior 179-80
PEN 73, 75 genius xxvii, xxviii, 449, 450, 452, 455,
personality dimensions xix, 17-33, 37 574
psychosis, study of xx, 118-19 creativity 404, 417-18, 419, 423^, 428,
reinforcement sensitivity theory (RST), 443, 450, 451, 452, 453, 455, 574
link to 38-9 domain specific selection 443-4, 454
scientific approach 18, 37 females 429
superfactors (P, E, N) 69, 87 genetics of 423-4
theory xvii, xx, 17, 68-70, 80, 109-12, heritability 424
217, 464-5 Kroeber, A. L. and emulation 493-4
Eysenck and psychology 31-3,568-70,575 P dimension and 22, 23
contribution and influence 133, 570 problems of 423-4
Freud and 32 psychopathology 430
natural science 569, 584, 585 state of 451, 453
reductionism xviii, 31, 575, 576, 583 graphology 511, 513, 514-19, 534, 574
Eysenck and psychopathology 364-80,381-2 Eysenck's early studies 514-19
contribution to 373 Eysenck's later studies 516-17
psychotic disorders research 374-5 status of conclusions 517-19
three-dimension theory 374-8, 380 Greeks
two-dimensional theory 367-74, 380 humor 12-13, 36
Eysenck as teacher and mentor 543-58 Grossarth-Maticek, R. xx, 92, 93, 95-8,
"at homes" 554, 556 101, 335
character 545, 546, 547-50 evaluation of work 98-100
influence of 556-8 intervention studies 97-8
mentor, as a 555 investigative techniques 102
speaking style 552, 554-5 rationality/antiemotionality interview
teacher, as a 553-5 schedule 95
writer, as a 550-3 vitamin/mineral supplementation
writing style 551-2 research 104-5
614 Subject index

heliobiology 492, 493, 503, 507 learning theory 3

Chizhevsky, A. L. 492-3, 503 intelligence 31, 144, 145, 229, 462
solar activity 503-7 Abecedarian Project xxvi, xxvii, 312-13,
sunspots 503, 504 324, 325
hormone levels and behavior age and xxvii, 315, 316, 317-18, 322
androgen 181, 182, 183 American 264, 265
creativity and 438-42 American Asians 269, 276
estrogen 182, 183 American ethnic Mongoloids 263,268-9
human ability Australians 276
mental speed 285 biological factors 244-8, 247
noncognitive factors 284-5 children 312-14
human nature Chinese 264, 265
methodology 581-3 civilizations development 262
molecular vunerability 580 conceptual aspects of (A, B, C) 224,
natural science 583-5 225, 314, 320
study of 563-87 creativity and 406-7, 417
twenty-first century science 576-86 cultural variations 315
uniqueness and causality 579-81 deaf people 313-14
DNA markers 251-2
impulsivity xxiixiii, 49, 114-15, 152, 153, East Asian Mongoloids 262-8
248, 249, 345, 371, 573 environmental influences 315
arousal and 190, 198-203, 573 evolution of racial differences 274-7
Attention Deficit-Hyperactivity Eysenck's work 282-7
Disorder 189, 198 Galton-Spearman-Eysenck school 220,
behavioral approach system (BAS) 221, 224-6
153, 193 genes, specific 251
caffeine, effect of 202, 203 genetic and biological basis xxiii, xxiv,
children and adults 198 xxv
classical conditioning studies and genetic basis to racial differences 272-4,
200-1, 343-4 278
cognitive performance 201-3 genetics xxvi-xxvii, 240-4, 251, 260, 269
components of 191, 196, 200 geographical variation xxiii, xxiv-xxv,
conditioning and 200-1 259-79
description of 190 heritability xxiv, 240-2, 249-51
extraversionxxii, 190,191,192-3,197,201 impulsivity and 203
Eysenck and 190 infancy prediction 319-20, 321
impulsiveness scale (I7) intellectual genotypes 314-16, 319, 323
location in personality models xxu'-xtiii, Intelligence journal 287, 288, 292
114-15, 197, 204 Japanese 264, 265, 266
measurements of 194-8 Korean 266-7
PEN model and 193-4, 199 mental speed 284
sensation seeking 200 neural plasticity hypothesis 440-1, 442,
sensitivity 201 449, 451
time of day, effect of 202 Nufferno Test 217-18
inhibition 389 nutrition xx, 572
Eysenck hypothesis 395 personality and xxiii, 248-9
Hull concept 340, 341, 346 phenotype change and genotypes 318-19
 Subject index 615

phenotype correlations 243, 572 Differential Ability Scale xviii, 269

Roth research 226, 227, 228, 234 292 ECTs 236
specificity doctrine 222 inspection time (IT) xxvi, 225, 303
scientific approach 571 Metropolitan Achievement Test 325
sub-Saharan Africa 269-72 movement time (MT) 298
Taiwanese 265, 267 Multidimensional Aptitude Battery
true levels 316-17 (MAB) 249
types of 240 nerve conduction velocity (NCV) xxiv,
intelligence and information processing 244, 247-8
xxiii, xxiv, xxvi, 282-305, 573 odd-man-out discrimination task 235
combined study approach 289 overlap model 317-18, 319
criticism of combined approach 289-90 quantitative trait loci (QTL) xxiv, 251-
EEG and evoked potential correlates 2,253
xxvi, 293-4 reaction time (RT) xxiv, xxvi, 219-20,
Eysenck 282-7, 291, 292 225, 226-8, 229, 242-4, 262, 303
inspection time 300-3 spatial ability 264, 265, 275, 276
parallel processing 287-8 Standard Progressive Matrices 268, 269,
reaction time paradigms xxiv, xxvi, 297, 271, 272
299, 300 string measure 245
research, changing face of 287-91, 303 true-score variance 222-3
research, future 304-5 Wechsler Adult Intelligence Scale
research, present day 293-4 (WAIS) 220
Sternberg, R. J., componential Wechsler Intelligence Scale for
approach 294-6, 300, 303 Children-Revised (WISC-R) 241,
intelligence and malleability xxiii, xxvi, 263, 264
311-28 Wechsler Full Scale IQ 236
adoption 319, 323, 326 introversion (I) 19, 20, 31, 119, 143, 151,
effects of schooling 321-4, 322-3, 326 152, 154, 157, 159, 370
environmental influence creator 327-8 arousal and 9, 199, 200, 201, 342
gravitation hypothesis 327-8 conditioning and 147, 153, 343, 353,
naturally occurring changes 321 358
occupational status 327-8 cortical arousal 390
predictability 320-1 creativity 438
preschool intervention 324-6 drugs and 4, 371
variation in similar environments dopamine activity (DA) 398
312-14, 573 motor expression 394-6, 467, 468
intelligence testing 225, 264-5 neuroticism, effect of 470-1
averaged evoked potentials (AEP) xxiv, reward and punishment 481-2, 483-5
220, 225, 238, 244-6, 303 sensory psychophysiological evidence
Berkeley Growth Study 318 393-4
Binet-Simon test xxiii, 221-2, 286 sexual behavior 167, 172, 180, 181, 183,
block design test 264, 287 184
Canadian Cognitive Abilities Test social anxiety 60
(CCAT) 266 stimulation sensitivity 388, 390, 392-3,
cerebral glucose metabolic rate 398, 573
(CGMR) xxiv, 244, 246-7 organizations 484-5
decision time (DT) 298 teaching methods 475
616 Subject index

introversion (I) (continued) reaction time (RT) xxvi, 227, 231, 234,
television and music, effect of 479, 480 242-4, 245
traits of 192, 391, 392 Wechsler Full Scale IQ 236
vocational preferences 465, 466 Wechsler Intelligence Scale for
work distraction 478-9 Children (WISC) 245, 271
introversion-extraversion (I-E) 143,367,370
IQ xviii Jensen, A. R.
adoption 319, 323 Hick paradigm study 297-9
Abecedarian Project 312-13, 324, 324 Hick study, criticism of 299
age and 315, 316, 322-3 job satisfaction 469-72, 476-8
Andersons's theory of uncorrelated genetic determinants of 480-1
increments 318 reward and punishment 481-5
averaged evoked potential (AEP) xxiv, Jung, C. G. 19
220, 225, 238, 244-6, 303
black-white difference 260-1, 270 Laing, R. D.
blue-collar workers 439 psychopathology and 364, 365
Caucasiods 263, 265, 272, 273, 275, 277,
278 MaCrae and Costa see Costa and MaCrae
Chinese 263, 265 memory
creativity 407, 409, 416-17, 428, 434-5, long-term (LTM) 232, 233
436-44 short-term (STM) 232, 233, 243
crime 29 Milwaukee Project 324—5
DNA markers 251-2 molecular creativity and ND4M
heritability 241, 250 dynamics of ND4M model 451-3
Japanese 265, 266 human nature and society 445-6
Korean 266-7 molecular model (ND4M) for
mental speed 284 444-5,446,447,449,450,451-3,454,
Mongoloids 265, 272, 274, 275, 277, 278 455
Negroids 272, 274, 277, 278 multidimensionality 447
nerve conduction velocity (NCV) and multiplicativity 447-9, 454
reaction time (RT) 247-8 nonlinearity 453-4, 454
occupational status 327-8 motor response assessment
predictability 320-1 contingent negative variation (CNV)
social classes and 327 396
South African study 269-72 dopamine activity (DA) 397-8
Taiwanese 265 Hoffman reflex (H-reflex) 396
test score variance 293 startle reflex 55-8, 60-2, 396
vitamin/mineral supplementation and
104 N dimension 3, 4, 17-20, 92, 95, 60-1, 70,
IQ and personality 102, 152, 154, 159, 160, 167, 178, 197,
effect on dimensions 26-7, 29 367, 379, 380, 571
IQ testing 18, 223, 230, 296, 304 visceral brain, role of 4
cognitive ability 224, 225 negative emotionality
Hick paradigm 298, 299, 300 Eysenck's Personality Questionnaire-
inspection time 301-3 Revised (EPQ-R) 119
quantitative trait loci (QTL) xxiv, 251- startle potentiation 62
2,253 neuroimaging xvii, 13
 Subject index 617

neurophysiological research 347-9 Pavlov xviii, 36, 37, 69, 76, 80, 123, 144,
brain pathways 11 146, 159, 340, 340, 347, 353-4, 571
circuit analysis 340 cortex, model of 340
theory, development of 12 influence on Eysenck xviii, 4, 36, 37
neuroticism 17, 31, 68, 70, 71, 72, 74, 87, introversion-extraversion 143
103, 124, 142, 157, 182, 192, 342, 370, theory of central nervous system
373 properties xix, 75, 389
accidents and 473 theory of nervous types (Russian
astrology 520 theory) 369-70
extraversion, effect on 470-1 Pavlovian Temperament Survey (PTS)
graphology study 516 xix, 79, 80, 81
impulsivity and 200, 201 personality xx, xxii, 4, 462
job satisfaction and 476-8 accidents and 472-4
N dimension see separate entry alternative-5 model xvii, 7, 197, 571
reward and punishment 482-3 arousal theory xvii, 388
sexual activity and 165, 171, 172, 173, big-3 model 7
174, 176, 183 big-5 model xviii, 7, 23, 24, 25, 31, 110,
vocational preferences 465, 466, 469 197
work performance indicator 467-8 big-6 model xviii, 21, 28, 30, 571
biological basis 36, 193
organizations businessmen in different vocations
extravert 484, 485 465-6
introvert 484-5 cancer-prone traits 95, 96, 101, 102
classical conditioning, role of xxvii, 339
P dimension xviii, xxi, 4,6,20-4,27,39,69, creativity and 406, 410-12, 435, 437
109,152,154,155,157,160,167,178, crime, relationship with 143
334, 379, 380, 429, 571, 572 criminal behavior traits 155
core of 5 criminal blame and 156-9
evolution of xx, 21 differentials 175-6
masculinity and 169 dimensions xvii, xxi, 28-9, 142, 145-6
problems with 22-3 disorders 114
psychopathy 376 distractibility at work 478-80
scale 378, 379, 451 Eysenck's Personality Questionnaire
sexual behavior 171, 172, 174, 176 EPQ see personality testing
parapsychology 526-36 Eysenck's theory of xvii, xx, 17, 68-70,
ESP 529, 530, 533 80, 109-12, 217, 464-5
Eysenck 492, 511, 526-9, 529-30, 532, Galen-Wundt-Eysenck circle 18, 19
574 genetics and 7-8
fraud and methodological evidence genotype-environment correlations xvii,
532-4 12
ganzfeld 527, 529, 531, 534, 535 gigantic-3 versus big-6 models xviii, 24-5
prerecognition 529 graphology and 516-17
psi 527,528,529,530,531, 532,533,574 Gray's model 193
psychic readers 530-1 hormones 181-3, 438-44
selection of evidence 531-2 impulsivity xxii-xxiii, 114-15, 197, 204
variable effect size 527 job preference and satisfaction 469-72
passive avoidence learning 154 marital happiness predictors 178
618 Subject index

personality (continued) 467, 470, 471, 472, 479, 485, 520

marital similarity 174-5 Eysenck's Personality Questionnaire
neurochemical brain pathways xvii, 11 (EPQ) 92, 109, 113, 114, 117, 143,
neurophysiological research 347-9 157, 159,160, 165-7, 168, 171, 178,
P dimension xviii, 4, 5, 20-4, 334 193,197,199, 337, 343, 375, 376, 379,
Pavlov, influence of 4, 37 409, 410, 471, 476, 485, 516
predicting job success 468-9 Eysenck's Personality Questionnaire-
psychobiological basis 3-13 Revised (EPQ-R) Polish version
psychoticism, as a dimension of xx, 77-9, 80, 82, 87
109-20, 379, 407 Eysenck personality profiler (EPP) 485,
reinforcement sensitivity theory (RST) 574
xviii-xix, 38-63 Hoffman reflex (H-reflex) 396
scientific study of 17-18 I5 194, 201, 471-2
scientific approach xviii, 571 I7 52, 194, 195, 197, 200, 201
sensation seeking xvii, 74 Lie scale 167, 178
sexual attitudes and xxii, 167 Locke-Wallace Marital Adjustment
sexually deviant behavior 184 Test 175
structure 6-7 Maudsley Medical Questionnaire
theory, constituents of 110-11 (MMQ) 109
traits 414 Maudsley Personality Inventory (MPI)
traits and planetary positions 520 70, 75, 109, 191, 343, 470, 485
training and 474-6 NEO Five-Factor Inventory 83, 85, 86
work mood studies 469 Nufferno Test 217-18
personality and sex xxii, 165-84 Personality Research Form-E (PRF-E)
Kinsey 165 249
Masters and Johnson 165 reaction time (RT) see separate entry
personality disorders 114 Social and Political Attitude Inventory 471
dynamics of 372-3 Spielberger's Trait Anxiety scale 49, 53
Einheitpsychose theory 374, 375 startle reflex 55-8, 60-2, 396
hysteria 368-9, 370, 374 Strelau Temperament Inventory 75
manic depression 374, 375, 427 physicology 577
psychosis proneness 378 all-bottom approach 577
schizophrenia 118, 119, 374, 375, 376, intersystemic 578
377, 380, 381, 407, 408-9, 427, 429, intrasystemic 578
430, 450, 451 methodology 581-3
schizotypy and psychoticism 378-80 psychology and, use of theory 577-8
three-dimension theory 374-8 reductionism xviii, 583
personality testing 116-17, 152 uniqueness and 580-1
Achenbach scale 129-30 positron emission tomography (PET) 246
antisocial behavior questionnaire psychiatry
(ASB) 113 DSM-III 365, 366
Blame Attribution Inventory 156, 157 DSM-IV 366, 368
complementary-trait effect 49, 53, 55 Eysenck, criticism of 366, 367
contingent negative variation (CNV) Eysenckian school 367
396 Kraepelinian binary model 374, 375,
Eysenck Personality Inventory (EPI) 379, 568
70, 75, 109, 192, 193, 197, 199, 343, new wave 365
 Subject index 619

psychology antisocial behavior questionnaire

abstraction 565, 566 (ASB) 113
all-bottom analyses 566, 568, 577 astrology 520
Aristotle 565 children and, 112-13
bottom-up analysis 567 conditioning studies 344
Democrit 565, 566 creativity 112, 117, 404, 407-8, 409-10,
differential 215 413, 415, 416, 429, 449-51, 574
early ideas 565 criterion analysis 111
Eysenck and 31-3, 133, 568-70 EPQ scale of 49
importance of measurement 214 impulsivity, component of 190,194,199
nature-nuture models 567-8 job satisfaction 477
organizational, research of 485-7 learning and 476
Plato 565, 566, 576, 583 P dimension 4, 5, 6, 20-24, 27, 39,
psychometrics 221 69, 109, 154, 155, 157, 160, 167,
science, as a 563-87, 584, 585 334, 429
surface analysis 566, 567 predisposition 4
top-down analyses 567 psychosis and 118-19
psychology and medicine 92-106 role of in crime 138, 149-50
cancer xx, 93-5, 96, 101-2, 103 RST and 48
intervention studies 97-8 scales 375
Grossarth-Maticek, R., evaluation of schizophrenia 118, 119, 573
work 98-100 scientific personality 410-16
vitamin/mineral supplementation 103-6 studies of 109-10
psychopathology therapy 113-14
cognitive dimension 372, 382 traits of 119-20
dynamics of disorder 372-3
dysthymic disorder 368, 370 race and intelligence xxv, 572
Eysenck's contribution to 364-80, 381-2 ability patterns 261-2
future directions 381-2 adoption 272, 273, 279
genius 430 American Caucasoids 263
hysteria 368-9, 370 black-white difference 260-1, 270, 273,
P, significance of 375-8 278
psychopathy 157-9, 189, 190, 345, 376, brain size 262, 272, 273^1, 279
369 civilization development 262
psychosis proneness 378 climatic predictability theory xxv, 275,
schizotyp and psychoticism 378-80 278, 573
three-dimension theory 374-8, 380 cold winter theory xxv, 275, 277, 278
two-dimensional theory 367-74, 380 East Asian Mongoloids 262-8
zone analysis 370 environmental theory 272-3, 274
psychopharmacology 10-12 evolution of differences 274-6
psychophysiological Eysenck's work 259-61, 278
motor response and extraversion 396-8 geographical variation 261, 272
traits 10 Miller's paternal provisioning theory
psychoticism xx, 70, 71, 85, 86, 87, 124, 277
142, 143, 379 racial ability patterns 261-2
addiction 115-16 racial hybrids 270-1
advantages of 117-18 reaction times 262
620 Subject index

reaction time (RT) xxiv, xxvi, 229, 242-4 Sargent, C. 335

Hick paradigm 227, 228, 230, 233, 234, scientists
235, 236, 238, 242, 286, 292, 297- genius 418-19
300, 303, 304 natural 431, 432, 433, 438, 439, 451
measurement controversies 232-3 personality 410-12, 435, 437
non-g factor 235-7 professors, studies of 412-16
odd-man-out discrimination task 235, psychologists, studies of 411, 412
236, 238, 242 Roe, A., eminent scientists study 431-4
psychometric g and 219-20, 225, 226-8, social 431, 432-4, 438, 439, 451
230, 231, 233, 238 sensation seeking xix, 10,74,153,198,200,
Roth research 226, 227, 228, 234, 292 371
Sternberg memory scanning 297 evoked potential (EP) and 9
working memory 232, 233 heritability of 7-8
reaction time-IQ correlation 242 scale (SSS) xvii, 5, 6, 7, 10, 52, 87, 358
bottom-up explanation 237-8 sexual behavior
nerve conduction velocity (NCV) and attitudes and personality 167
247 attitude areas 166
phenotypic correlations 243 conditioning and arousal 181
top-down explanation 237-8 criminality xxii, 171
reductionism 283 deviance 172-3, 180, 184
Eysenck and xviii, 31, 575, 576, 583 dysfunction 173-4, 184
reinforcement sensitivity theory (RST) Eysenck personality questionnaire
xviii-xix, 38-63 (EPQ) 165-7
behavioral activation system (BAS) 38, genetic factors xxii, 179-80
39, 40, 41, 42, 49, 51, 52, 53, 56, 57, Lie scale 167, 171, 175, 178
58, 60 153, 193 male-female differences 168-71
behavioral inhibition system (BIS) 38, marital satisfaction 174-5, 177
39, 40, 49, 53, 56, 57, 58, 60, 153, 193 personality and hormones 181-3
BIS-BAS interaction 40 personality characteristics xxii, 172
BIS-BAS questionnaire 52 social attitudes 176-7
Card-Arranging Reward socialization 123, 124, 148, 160
Responsiveness Objective Test passive avoidence learning 154
(CARROT) 51, 55, 58 Spearman, C. E. 215, 216, 219, 219, 220,
description of 38 222, 284, 286, 544
Geller-Seifter conflict task 47, 58-9 classical test theory 216, 221
link to Eysenck's personality theory factor analysis 216, 218-19, 224
38-9 intelligence factor g xviii, xxiii, xxiv,
procedural learning task 45, 46, 48-9 xxvii, 18, 21, 24, 216, 217, 218, 220,
Q task 44-5, 46 224, 225, 226-8, 237, 263, 265, 267,
response activation and behavioral 284, 285, 296, 302, 317, 438,448, 452,
inhibition (maze task) 46, 49-50, 55 558, 572
results, diversity of 62-3 Sternberg, R. J.
sex (gender), effects of 50-2 cognitive components approach 294
startle reflex 55-8, 60-2 intelligence testing 294-6, 300, 303
testing 41-4 memory scanning reaction time
variations in trait measures 52-3 procedure 297, 303, 304
 Subject index 621

temperament xix scales and PEN xix, 79-81, 82-4, 85, 86,
background to 68 87
behavior-genetic theory of 73-4 Thorndike, E. L. 543, 547, 551
dimensions, relationship to xix, 71-3 training
HAS Temperament Survey (EAS-TS) applications for personality variables
79, 80, 81 474-6
Eysenck theory of extraversion and aptitude batteries 475
neuroticism 68-70, 371 EPQ as a predictor of success 476
Heymans model 190 teaching method and personality 475
Hippocrates-Galen 38, 75
humor 12-13 variable, effect size 513-4, 527
interactional theory 74
location of PEN 82-6 Wundt, W. 215, 216
Pavlovian Temperament Survey (PTS)
xix, 79, 80, 81 Zuckerman, M. 37, 153, 169, 198, 200,
PEN personality factors and 70, 77 358, 371, 449, 571
regulative theory of, (RTT) xix, 76, 79, Sensation Seeking Scale (SSS) xvii, 5, 6,
80,81 7, 10, 52, 87, 358
Rusalov's theory of xix, 76, 79, 80, sensation seeking theory xix, 74, 79, 80,
81 81